AI Creators

Uzumasa JidAI Short Film Competition 2026: GMI Cloud Production Support | Enter by Nov. 15

aratama 璞 | Editor-in-Chief, AI Creators — Tue, 13 Oct 2026 01:19:35 +0000

Bring your own vision of history to life with AI. The Uzumasa JidAI Short Film Competition 2026 is accepting entries until Sunday, November 15, 2026.

The competition welcomes short films approximately 15–180 seconds long that explore history using generative AI. Creators are free to choose their AI tools, and films with watermarks are eligible. Winning entries are scheduled to be screened on Sunday, December 6, the final day of the Uzumasa Joraku Festival, with commentary from the judges.

Competition sponsor GMI Cloud is offering $30 in production credits to each of the first 30 eligible applicants creating a film for the competition. Read on for the credit application process, entry instructions, this year’s judges and four award-winning films from 2025 to inspire your next project.

At a glance

Submission deadline: November 15 at 11:59 p.m. Films should be approximately 15–180 seconds long.
Creative tools: Use any AI tools you like. Watermarked films are accepted.
GMI Cloud support: $30 in credits each for the first 30 eligible applicants. Activate by November 15; use by November 30.
Credit applications: Post on X with the required hashtags, then submit your post URL and other details through the Google Form.
How to enter your film: Follow the Uzumasa Joraku Festival’s official X account and quote-post its call for entries with your film.

Applying for production credits and submitting a film are separate processes. Credits must be activated by November 15 and used by November 30. Film submissions close on November 15 at 11:59 p.m.

History meets generative AI: share your vision in Uzumasa

The Uzumasa JidAI Short Film Competition brings together historical storytelling and generative AI. How do you see the past—and what kind of story or visual world would you build around it? The competition invites creators to explore history and period drama through their own creative perspectives.

Previous winners range from black-and-white period pieces and dreamlike action sequences to nostalgic portraits of Kyoto and playful takes on history. Your starting point could be the visual style, music or storytelling approach you already love working with.

Competition overview and key dates

Competition	Uzumasa JidAI Short Film Competition 2026
Theme	History-themed short films created using generative AI
Length	Approximately 15–180 seconds
Tools	Any AI tools; watermarked films accepted
Submission period	October 7–November 15, 2026, closing at 11:59 p.m.
Judging period	November 16–28, 2026
Winners announced	December 6, 2026. Screenings and judges’ commentary are planned for the final day of the Uzumasa Joraku Festival.
Awards	One Grand Prize and two Outstanding Awards, according to the official call for entries on X
Judges	Sissy, Holly, Genki Takigawa and Norihiro Yamaguchi
Sponsors	GMI Cloud and CapCut
Official website	Uzumasa JidAI Short Film Competition

Sources: Official entry guidelines / Competition announcement / Official call for entries on X

GMI Cloud offers $30 in production credits to the first 30 eligible applicants

As a competition sponsor, GMI Cloud is supporting creators working with generative AI. The first 30 eligible applicants creating a film for this competition can receive $30 in GMI Cloud credits each.

Use the credits to develop your entry, from experimenting with historical settings to bringing characters and scenes to life. Credits must be activated by November 15, 2026, and used by November 30, 2026.

Apply for production credits through the Google Form

The application process has been updated. Post on X using the required hashtags, then enter your post URL and other details in the Google Form. After your application is reviewed, you will receive your credit activation code by email.

This replaces the previously announced process using direct messages on X. Posting on X alone does not complete your application: you must also submit the Google Form.

Support	$30 in GMI Cloud credits per creator
Eligibility and availability	The first 30 eligible applicants creating a film for this competition
Activation deadline	November 15, 2026
Use-by date	November 30, 2026
How to apply	Post on X with the required hashtags, then submit your post URL and other details through the Google Form
How credits are delivered	An activation code is sent by email after your application is reviewed

How to apply for production credits

Post your credit request on X using the required hashtags.
Include all three hashtags: #太秦歴史AIショート #UzumasaJidAIshort #GMICloud. Copy the URL of your post so you can enter it in the form.
Complete and submit the Google Form.
Open the credit application form and enter your family name, given name, email address, X profile URL, the URL of your credit request post and the AI models you currently use. Answer every question and review the notices in the form before submitting.
Follow the email instructions to activate your credits.
After your application is reviewed, you will receive a credit activation code by email. Follow the instructions and activate your credits by November 15. If you do not yet have a GMI Cloud account, you can create one here.
Create your film and enter the competition.
Use your credits to make your film, then submit it following the instructions from the Uzumasa Joraku Festival’s official X account. Applying for credits and entering your film are separate processes.

For updates, follow both official GMI Cloud accounts on X: the global account (@gmi_cloud) and the Japan account (@gmicloudJapan).

Applications with missing or incomplete answers are not eligible. Credits will also be withheld if GMI Cloud determines that they would be used for purposes other than creating a competition entry. Check the official announcement and application form for current availability and the latest conditions.

Apply for credits via Google Form
View GMI Cloud’s application announcement

Sources: GMI Cloud credit application form / Official application announcement on X

Enter the competition on X in two steps

Submit your film through the Uzumasa Joraku Festival’s official X account (@Joraku_1059). Even if you have applied for GMI Cloud credits, you must still enter your film separately.

1. Follow the Uzumasa Joraku Festival on X

Follow @Joraku_1059. The official guidelines require this so the organizers can contact award winners.

2. Quote-post the call for entries with your film or a viewing link

Quote-post the official call for entries and include your film. If you cannot attach the video directly on X, you can submit a link to a film hosted on YouTube or another platform.

Use a link that allows people to watch without logging in.
Keep your film available to view until the winners are announced.
Include the hashtags specified on the official X account: #太秦歴史AIショート and #UzumasaJidAIshort.

View the official call for entries

Meet the four judges: perspectives from AI filmmaking and period cinema

The 2026 panel brings together AI filmmakers and specialists in film and Japanese period drama. Sissy and Holly, both Outstanding Award winners in 2025, return this year as judges.

Sissy | AI filmmaker

Sissy creates imaginative, story-driven films exploring fantasy, mythology and renewal. Their film Genmu Uzumasa received an Outstanding Award in the 2025 competition. Sissy is also listed on AI Creators, where you can explore their work and portfolio under the profile name Shisi.

en.ai-creators.tech

Shisi

https://en.ai-creators.tech/creators/shisi/

An AI Filmmaker Crafting Fantastical Worlds of Myth and Renewal. Since beginning AI video production in December 2024, this creator has developed a distinctive visual style centered on fantasy, mythology, and renewal.

Holly | Film director and AI creator

Holly studied film criticism at Kyoto University and now works in AI-assisted filmmaking. Their projects include the fully AI-generated animated feature Shichigatsu no Android. In 2025, An Untold Tale of Sawako Tange: The Million-Ryō Pot received an Outstanding Award in this competition.

Genki Takigawa | Director, screenwriter and producer

A graduate of Ritsumeikan University’s College of Image Arts and Sciences, Takigawa works across film development, producing, screenwriting and directing. His credits include Oni Girl!! and the Battle King!! series.

Norihiro Yamaguchi | Period-film scholar and former president of Toei Kyoto Studio Park

Yamaguchi spent many years working in character-based entertainment at Toei’s events division. His research focuses on Japanese period cinema, including preserving and passing on studio techniques and film materials. The competition’s official X account identifies him as chair of the judging panel.

Sources: Official judge profiles / Judging-panel chair announcement on X

Watch the 2025 award winners

There is no single way to tell a story about history with AI. These four award-winning films, featured on the competition’s official website, offer very different approaches. Watch their visual choices, pacing and structure—and consider where your own ideas could take you.

The descriptions below are adapted from the judges’ comments on the 2025 entries. They do not represent the judging criteria for 2026. If an embedded video does not play, use the YouTube link beneath it.

Genmu Uzumasa | Sissy — Outstanding Award

Genmu Uzumasa by Sissy | Watch on YouTube

Oiran courtesans, samurai and ninja appear in a dreamlike world built from period-film imagery. The official comments praise the distinctive creative vision that unifies its varied scenes, as well as its visual rhythm and polish.

An Untold Tale of Sawako Tange: The Million-Ryō Pot | Holly — Outstanding Award

An Untold Tale of Sawako Tange: The Million-Ryō Pot by Holly | Watch on YouTube

Black-and-white imagery evokes the atmosphere of classic Japanese period cinema. The judges’ comments highlight the characters’ presence and sword-fighting sequences, along with the humor woven into its otherwise serious tone.

Echoes of Uzumasa | Konoha — Outstanding Award

Echoes of Uzumasa by Konoha | Watch on YouTube

Music and evocative landscapes recall Kyoto and the film studios of Uzumasa. The official comments praise the film’s sense of nostalgia and its portrayal of an elegance passed down through the ages.

We Make History | Eiji Ise — Special Award

We Make History by Eiji Ise | Watch on YouTube

This playful film pairs history with an unexpected dose of cuteness. The judges’ comments praise its cheerful entertainment value and its idea of showing characters crawling like babies, regardless of their era or social standing.

Films and judges’ comments: Official 2025 winners archive

Before you submit: originality, music and usage rights

Read the official entry guidelines and rules in full before submitting. Key requirements include:

Entries must be entirely original. Films that infringe another person’s rights are not eligible.
Production methods that imitate a particular well-known artist’s style are prohibited. The official rules give the example of using an artist’s name in a prompt.
Use music and audio you created yourself, or obtain permission covering both public screenings and publication on social media.
Entries must not violate public order or accepted standards of decency.
Under the rules, entrants retain copyright in their films. The organizers may, however, use and edit submitted films without payment for event-related purposes, including screenings, social media and publicity.
Entrants are responsible for resolving any disputes arising from their films’ content.

Before publishing or submitting your film, also check the terms of the generative AI services and audio assets you use.

Read the official entry guidelines and rules

Related events and production support for students

October 19: award winners share their process at the Samurai AI Filmmaking Symposium

The Samurai AI Filmmaking Symposium takes place on Monday, October 19, from 4:30 to 6:00 p.m. in the conference room of the Hirai Kaichiro Memorial Library at Ritsumeikan University’s Kinugasa Campus. Holly and Sissy are scheduled to discuss their creative processes and techniques, with examples from their work.

Admission is free, with advance registration required and capacity limited to 50 attendees. According to the official X announcement, judging-panel chair Norihiro Yamaguchi will also give a roughly 15-minute overview of the competition and its key points, starting at 5:45 p.m.

View symposium details and register

Sources: KYOTO CMEX event listing / Additional details on the official X account

CapCut production support for students

CapCut is also offering production support for students who wish to enter the competition: seven days of CapCut Pro access and 5,000 AI credits. Check the official announcement for eligibility and application requirements. Applying for this support and entering the competition are separate processes.

Explore CapCut’s student production support

Submit by November 15—and bring your own vision of history to the screen

A dreamlike period drama, a story with a sense of humor, or a historical landscape shaped by music: past winners show just how much room there is for different approaches. Start with what you do best, and build it into a film of approximately 15–180 seconds.

Entries close on Sunday, November 15, 2026, at 11:59 p.m. If you would like to apply for GMI Cloud production credits, check availability and the application conditions early enough to leave time for your film.

View entry guidelines
Apply for GMI Cloud credits

Related reading: how GMI Cloud supports AI creators

GMI Cloud also hosts events where creators can showcase AI-generated work and connect with one another. Our earlier article introduces the event held at Shibuya Hikarie on October 8.

GMI & MiniMax AI Art Gallery: event overview and call for submissions

This article was prepared by AI Creators using the competition’s official website and X account, materials supplied by GMI Cloud and other cited sources. Information is current as of October 13, 2026. Competition details and the availability of production support may change. Check the official announcements for the latest information.

The post Uzumasa JidAI Short Film Competition 2026: GMI Cloud Production Support | Enter by Nov. 15 first appeared on AI Creators.

AI Greta Garbo: What SKF’s Ad Means for Likeness Licensing

aratama 璞 | Editor-in-Chief, AI Creators — Mon, 05 Oct 2026 08:49:51 +0000

When AI is used to bring a deceased star back into advertising, what exactly needs to be licensed before a brand can use “the person”?

On September 30, 2026, SKF, the Swedish industrial technology company best known for its bearings, launched a global campaign featuring an AI recreation of Greta Garbo, the Swedish-born Hollywood star of the 1920s and ’30s. Garbo died in 1990, so the campaign brings her back to the screen 36 years after her death. According to The Guardian, Garbo’s likeness had appeared posthumously in fashion campaigns for Ferragamo, Gucci, and Louis Vuitton, but this is the first time it has been used in moving images.

What makes the project interesting is how it was made. According to press reports, the production team did not feed copyrighted photographs or film footage of Garbo into the AI models used to generate her appearance. Her voice, however, was recreated from recordings of her early films. And SKF states in its press release that the project was developed in collaboration with Garbo’s family and Harriet Brown & Company Inc.

If AI can recreate a person’s appearance without ingesting protected photos or footage, what is left to license? This case offers a useful lens on how identity itself may need to be licensed in the AI era.

Key takeaway

SKF’s campaign does not suggest that a brand can avoid permission simply by keeping photos and footage out of the AI pipeline.

The more generative AI can reconstruct a recognizable person without ingesting protected source material, the more the central question shifts beyond copyright to who controls, and who can license, the personal and commercial value of a person’s likeness, voice, name, and legacy.

SKF’s Campaign and Its Connection to Garbo

SKF announced the project in a press release titled “Greta Garbo is back.” The campaign runs under the concept “Pay Like It’s 1919” and features Garbo introducing SKF’s magnetic bearings, which the company says eliminate friction entirely by suspending a rotating shaft in a magnetic field.

The “1919” refers to a payment model SKF says it introduced that year, under which customers could finance new equipment through the energy savings it generated.

Garbo’s own connection to SKF dates from a little later. Before becoming an international star, she appeared around 1921 in a promotional film presenting Swedish industry to overseas audiences, and SKF was among the companies featured.

In other words, 1919 is the origin of SKF’s payment model, while around 1921 is when Garbo first appeared in a film promoting the company.

Primary source: SKF press release: “Greta Garbo is back – starring in SKF’s initiative to overcome friction”

SKF has also released the campaign film.

What Was Kept Out of the AI Pipeline, and What Was Still Used?

To understand the production, it helps to distinguish between saying archival material was “not used” and saying that protected photos and footage were not fed into the models that generated her appearance.

According to the Swedish film publication Moviezine, old films, photographs, and recordings served as reference material for the production. What the press materials emphasize is that protected archival imagery and footage were not fed into AI to generate the new visuals. The Guardian likewise reported that, because the material is protected by copyright, the creators did not feed Garbo’s film footage or photographs into the AI. Her appearance was instead built mainly from text prompts describing her features.

Nor was the performance generated entirely by AI. According to The Guardian, a real actor performed the role, and an AI-generated Garbo appearance was layered over that performance.

The production, in short, combined:

Performance: a human actor
Appearance: generated from text prompts, without feeding protected photos or footage into the models
Voice: recreated from recordings of Garbo’s early films
Direction and editing: a human production team

As for the tools, The Guardian named ByteDance’s Seedream and Google’s Nano Banana Pro. Reuters added that ByteDance’s Seedance and Kuaishou’s Kling AI were used for motion, and that licensed audio from Garbo’s early films was used for the voice and was the only archival material fed into the AI systems. These production details do not appear in SKF’s press release; they come from press reporting.

Copyright and Identity Are Becoming Separate Layers

If a scene from one of Garbo’s films were reused in an ad, the main legal questions would be copyright in that film and permission to use it.

Generative AI changes the structure of the problem. A production team can now create someone who looks like Garbo without feeding the original footage into a model at all. The relevant rights and interests begin to separate into distinct layers:

copyright in existing photographs, films, and recordings
the person’s likeness
the person’s voice
their name
publicity value
trademark and unfair competition
posthumous personality interests and legacy management

Copying a work and recreating a person are becoming technically separable.

The distinction matters here because the visual and audio components were handled differently: the visuals reportedly avoided ingesting protected photos and footage, while the voice drew on existing film recordings.

Is Family Cooperation Enough?

SKF says the support of Garbo’s family and Harriet Brown & Company Inc. ensured that her legacy was handled responsibly. In that sense, the campaign is structurally different from an unauthorized deepfake.

But family cooperation does not automatically settle every legal question. Who holds which rights, and who can license them, depends on the person, the jurisdiction, the contracts involved, and the source material. Legal permission and cultural acceptance are also separate questions.

The Guardian‘s review was harsh, calling the ad dull and humorless and raising broader concerns about the use of celebrity likenesses through AI and the need for guardrails. By contrast, Craig Reisfield, one of three relatives who manage Garbo’s estate, praised the recreation as strikingly realistic and said he was open to using her in future film roles.

What Needs to Be Licensed? Designing a Digital Likeness Agreement

As brands use AI to recreate actors, celebrities, and other public figures, traditional advertising agreements may no longer be granular enough. A digital likeness agreement may need to address:

which age or period of the person’s appearance may be recreated
how far the appearance may be altered
whether the person’s voice may be generated
whether the digital version may say things the real person never said
which media are covered, including video, still images, social media, games, and interactive content
whether the person’s data may be used to train AI models
whether generated models or digital assets may be reused in other projects
what happens to models, source material, and training data when the agreement ends

Traditional ad agreements have focused on where, and for how long, a finished campaign can run. With generative AI, a digital version of a person can, in theory, produce an unlimited number of new performances.

That is why contracts may increasingly need to govern not only the use of a finished work, but the future capacity to generate new versions of the person.

If This Happened in Japan, Who Could Grant Permission?

For a Japan-based publication, the obvious next question is how this would play out under Japanese law.

On the same day SKF announced its Garbo campaign, the Tokyo District Court ruled in the Kenjiro Tsuda AI voice case. The court stated, as a general principle, that a person’s voice, like their portrait, is a symbol of the individual’s personality and may in certain circumstances be protected by the right of publicity. That points toward protection for living people. But what about the dead?

A report published by Japan’s Ministry of Justice in August 2026 says that there is currently no settled view on whether the right of publicity can be transferred or inherited, and that the question should be studied further in light of case law and other developments (p. 49).

The right not to have one’s portrait and similar attributes used without justification is a separate matter: the prevailing view is that this right lapses on death and does not pass to heirs.

Surviving family members may, however, have their own protectable interest when conduct infringes their feelings of reverence and affection for the deceased, and the report says the same framework can apply to the use of a deceased person’s voice (pp. 61–66).

Under the Unfair Competition Prevention Act, the report also suggests that where a talent agency or similar entity continues to manage the commercialization of a person’s portrait after death, that entity may be able to bring a claim (p. 71).

In other words, Japanese law supports neither a simple rule that “family permission is enough” nor the idea that a deceased person’s likeness is free for anyone to use. Who has authority to license a deceased person’s identity may itself have to be examined right by right and case by case.

Reference (in Japanese): Ministry of Justice, Japan – Full text of the study group report

AI Creators’ View: The Next IP May Be the Person, Not the Work

The core IP assets of the content industry have traditionally been works: films, music, photographs, illustrations, and characters. Generative AI changes that. As systems learn to reproduce a person’s appearance, voice, expressions, movement, and manner of speaking, the attributes of the person themselves can become reusable creative assets.

That does not mean the SKF campaign proves a market for posthumous digital likeness already exists. What we can say is narrower: a major brand has worked with a late film icon’s family and Harriet Brown & Company to use her digital likeness in a commercial campaign, and at least one relative who manages the estate has expressed openness to future uses.

Editor’s Note

AI can now recreate a recognizable person without feeding their photos or footage into a visual model. That does not make copyright irrelevant. It widens the question from who owns the work to who controls the commercial value of the person.

The Kenjiro Tsuda AI voice ruling in Japan and SKF’s Greta Garbo campaign point to the same underlying question: if resemblance itself has value, who does that value belong to?

Conclusion: “What Are We Licensing?” May Become the Central Question

SKF recreated Greta Garbo for a moving-image ad 36 years after her death. According to press reports, protected photos and footage were not fed into the models that generated her appearance; the production combined text-prompt generation with a human actor’s performance, while her voice was recreated from recordings of her early films. SKF says the project was developed with Garbo’s family and Harriet Brown & Company Inc.

The significance of the project is that rights clearance in the AI era can no longer be reduced to whether an original copyrighted work was copied.

It looks like her. It sounds like her. Audiences watch because it seems to be her.

Who controls that value, and who has the authority to license it? Designing the rights and contracts around digital likeness is set to become an increasingly important part of AI-driven creative production.

This article provides general information based on public reporting and government materials and is not legal advice on any specific matter.

Key References

The post AI Greta Garbo: What SKF’s Ad Means for Likeness Licensing first appeared on AI Creators.

AI Soundalikes and Publicity Rights: Japan’s Tsuda Ruling

aratama 璞 | Editor-in-Chief, AI Creators — Mon, 05 Oct 2026 08:30:38 +0000

When generative AI is used to create a voice that sounds like a real person, what should AI creators and companies be watching for?

On September 30, 2026, the Tokyo District Court (46th Civil Division) addressed that question in a case brought by Kenjiro Tsuda, a voice and screen actor known to anime fans as the voice of Kento Nanami in Jujutsu Kaisen, against TikTok Pte. Ltd., the Singapore company that operates TikTok. The court stated that a person’s voice, like their portrait, is a symbol of the individual’s personality, and may in certain circumstances be protected by the right of publicity. Kei Hirano, one of the plaintiff’s lawyers, said it was the first ruling to expressly recognize voice as falling within the right of publicity.

The court did not, however, find that the narration at issue actually infringed Tsuda’s rights. Because the posts and their data were deleted while the case was pending, it dismissed his claim for removal. The ruling does not draw a simple line that “making an AI voice that sounds like someone is illegal.”

Even so, it matters to anyone using AI-generated voices in content, advertising, social media, film, or games. Even if a person’s voice was never used as training data, legal issues can arise when the output evokes that specific individual and is used to exploit their customer appeal.

Key takeaway

The question is not only “Whose voice was used for training?” It is also: Who does the final output sound like? Does it trade on that person’s customer appeal? How is the voice presented, and how is it monetized?

The ruling is best read as opening the door to a broader debate over how the personal and economic value of a voice should be treated in the age of AI.

What Was at Issue in the Tsuda Case?

The case (Tokyo District Court, Case No. 2025 (Wa) 70635) was an action seeking the removal of posted content. At issue were 188 videos uploaded between July 2024 and September 2025 by an anonymous user’s TikTok account. Each paired narration with visuals and subtitles on what the judgment describes as “trivia on urban legends, paranormal phenomena, and the like.”

Tsuda argued that the narration resembled his voice and sought removal under either Japan’s Unfair Competition Prevention Act or the right of publicity.

A website run by the uploader, however, carried a statement to the effect that the narration was not made by training an AI on Tsuda’s voice without permission, but by training it on the voice of a friend who is good at voice impressions. The court recognized only that this statement appeared on the website; it made no finding that a friend’s voice was in fact used. The judgment also notes that comments on the videos included viewers saying the voice sounded like Tsuda, and others doubting the uploader’s explanation.

Primary source (in Japanese): Tokyo District Court, Judgment of September 30, 2026, Case No. 2025 (Wa) 70635

The Court’s General Principle: Voice Is a Symbol of Personality

The most closely watched part of the judgment was the court’s statement that a person’s voice, like their portrait, is a symbol of the individual’s personality.

Japan has no statute expressly codifying the right of publicity; the doctrine has developed through case law. The Supreme Court first addressed it squarely in its 2012 Pink Lady decision, describing it as the right to exclusively exploit the customer appeal of a person’s name, portrait, and similar attributes. The Tsuda ruling expressly places voice within that framework.

The court then set out a general test: the unauthorized use of, for example, a performer’s voice infringes the right of publicity where the voice has customer appeal that promotes the sale of goods or services, and the use is made for the sole purpose of exploiting that customer appeal.

The court did not, however, decide whether the narration in the 188 videos met that test. The significance of the ruling lies not in drawing an infringement line around specific audio, but in giving voice a legal foothold within the right of publicity.

Why Was the Removal Claim Dismissed?

The uploader deleted the account on May 29, 2026, while the case was pending. Because no request to restore it was made within 30 days, the court found that the account data, including the videos, had been deleted from TikTok’s servers as of the end of June 28, 2026. Tsuda argued that published data-retention periods do not always match actual practice, so it was unclear whether the data was truly gone, but the court rejected that argument.

TikTok had also argued that, because the videos no longer existed, Tsuda lacked a legal interest in bringing the suit and the case should be thrown out. The court rejected that argument too. Whether the content still existed, it held, was not a question of Tsuda’s interest in bringing the suit but a substantive question of whether a right to removal actually existed.

It then concluded that, even assuming TikTok could be subject to a removal obligation, there was nothing left to remove, and dismissed the claim on the merits.

The judgment states expressly that the court would not decide the remaining issues. It did not rule on Tsuda’s claims under the Unfair Competition Prevention Act, or on whether TikTok was under any obligation to remove the posts.

The Ministry of Justice Had Already Issued Guidance on Voice

On August 7, 2026, Japan’s Ministry of Justice published a report by its study group on civil liability for the unauthorized use of portraits, voices, and similar attributes, subtitled “Interpretive Guidelines on Infringement of the Right of Publicity by Generative AI” (English title ours). Prompted by concerns that the unauthorized use of portraits and voices through generative AI has become serious, the report sets out how existing law and case law apply to the right of publicity, the right not to have one’s portrait and similar attributes used without justification, and the Unfair Competition Prevention Act.

The report says that a voice both identifies a person and symbolizes their personality, and that there was no dissent within the study group that voice falls within the scope of the right of publicity. It also says that nationwide fame is not required for customer appeal: a voice can have it within a particular field or region.

Where the court’s ruling stayed at the level of general principle, this report offers more concrete guidance for people actually making and deploying AI voices.

Reference (in Japanese): Ministry of Justice, Japan: Publication of the study group report on civil liability for unauthorized use of portraits, voices, etc.

If the AI Never Trained on the Person, Can It Still Be a Problem?

For AI creators, the most striking detail in the case is the uploader’s claim that the AI was trained on a friend who is good at voice impressions. Even if that were true, would it matter if the output still sounded strikingly like Tsuda? The court did not decide that question, but the Ministry of Justice report offers a framework.

What matters is who listeners think is speaking

Where a voice resembles a real person, the report says the analysis should weigh the degree of similarity together with any additional cues that bring that person to mind, and ask whether customers would identify the voice as theirs (pp. 26–27). Scientific tools such as voiceprint analysis can be one factor, but they are not decisive (footnote 56).

The test, in other words, is not acoustic similarity alone. It is who the listener believes they are hearing.

Undisclosed impersonation can cross the line

The report notes that impersonators ordinarily perform under their own names, so impressions are generally not treated as use of the original person’s “voice” and do not infringe the right of publicity.

However, where the impersonator’s name is not disclosed, a voice resembling someone with customer appeal is used, and audiences may take it to be that person, there is room for infringement (p. 27). The same logic may apply to AI voices: how a voice was technically made may matter less than whose voice it is presented as.

Keeping the person out of the training data doesn’t eliminate the risk

On portraits, the report states that even where neither the training data nor the prompt includes the person’s actual likeness, a similar image can still be produced by adjusting prompts, and that using such an image for the sole purpose of exploiting the person’s customer appeal may infringe the right of publicity (footnote 194).

That passage concerns portraits, and no court has ruled on the point for voice. Still, the practical lesson is clear: “we didn’t train on that person” is not, on its own, an answer to every legal question about commercially using their identity.

Does Labeling It “AI Voice” Make It Safe?

Disclosures such as “This is not the real person” or “This voice is AI-generated” matter, but their legal effect depends on the right involved. In this case, TikTok argued that because the uploader’s website made clear the voice was not Tsuda’s, there could be no confusion. As noted above, the court did not decide that point.

The Ministry of Justice report explains that under Article 2(1)(i) of the Unfair Competition Prevention Act, which covers causing confusion with a well-known indication, a disclaimer, such as a note on photo merchandise or calendars stating that the product has no connection to the person pictured, may mean the confusion requirement is not met (p. 74). By contrast, Article 2(1)(ii), which covers misappropriating a famous indication, does not require confusion at all. Tsuda relied on both provisions.

The right of publicity raises a further question. It is not limited to whether viewers mistook the content for the real person; it also asks whether the use is made for the sole purpose of exploiting that person’s customer appeal.

Our editorial view, then, is that disclaimers can reduce the risk of confusion, but they should not be treated as a complete answer under either unfair competition law or the right of publicity.

Five Practical Checks for Anyone Creating AI Voices

1. Check not just the training source, but who the output sounds like

Even if the person’s own recordings were never used, the result may strongly evoke a specific individual. For commercial use, review where the data came from, how similar the final output is, and how it is presented.

2. Ask whether “sounding like them” is the selling point

If a voice draws views because it sounds like a particular celebrity, or lifts ad performance because audiences associate it with that person, the use may be trading on their customer appeal.

3. Look at how the voice is monetized

Ad revenue, platform payouts, product sales, customer acquisition, and promotions all raise the commercial stakes and call for closer scrutiny.

4. Don’t rely on disclosure alone

Labeling content as AI-generated is an important transparency measure, but it does not by itself resolve every rights issue.

5. If you are deliberately recreating a real person, build on permission

When a commercial AI voice is designed to evoke a real actor, voice actor, narrator, singer, influencer, or other identifiable person, the safer approach is to structure the project around permission from that person or the appropriate rights holder.

VTubers need a more nuanced look. The report notes that the voice of a fictional character, where it is not a real person’s voice, does not carry customer appeal protected by the right of publicity. But where a performer’s voice is identifiable as a real person’s and their customer appeal is being exploited, further analysis is needed.

AI Creators’ View: Voice Is Becoming an Identity Asset

AI voice has mostly been discussed as a technical question: how to generate convincing speech from data. Read alongside the Ministry of Justice report, however, the Tsuda ruling makes clear that a voice is more than sound. It carries personal and economic value, identifying an individual and conveying their reputation, popularity, and persona.

As AI makes it easier to produce a voice that sounds like someone without copying their recordings, the central question moves beyond training data. Who does the audience hear? Whose value is being used?

Editor’s Note

Rights management in the AI era can no longer stop at asking what went in. We also need to ask whose personal and economic value the final output invokes.

For AI creators, building that judgment into the production process is becoming part of the work itself.

Conclusion: It’s Not Only About Whose Data Was Used

The Tokyo District Court stated, as a general principle, that a person’s voice, like their portrait, is a symbol of the individual’s personality and may be protected by the right of publicity in certain circumstances. It did not find that the narration in this case infringed Tsuda’s rights, and it did not decide the Unfair Competition Prevention Act claims or whether TikTok had any obligation to remove the posts.

The practical issues for AI voice users are nonetheless coming into focus: training source, final output, similarity, presentation, purpose, monetization, and permission. Anyone using AI voices in commercial content should treat these as a single design question, not a checklist of separate ones.

This article provides general information based on the judgment and public materials and is not legal advice on any specific matter.

Key References

en.ai-creators.tech

Generative AI Creative Services for Business

https://en.ai-creators.tech/enterprise/

AI Creators supports companies with generative AI creative production, AI creator matching, corporate training, strategic advisory, PR, and partnerships.

The post AI Soundalikes and Publicity Rights: Japan’s Tsuda Ruling first appeared on AI Creators.

Thomson Reuters v. ROSS: Appeals Court Rejects AI Fair Use

aratama 璞 | Editor-in-Chief, AI Creators — Mon, 05 Oct 2026 01:50:48 +0000

For the first time, a U.S. federal court of appeals has ruled on whether using copyrighted material to train an AI system is fair use. In Thomson Reuters v. ROSS Intelligence, the answer was no.

In an opinion dated September 29, 2026, and released in full the following day, the U.S. Court of Appeals for the Third Circuit held that ROSS’s use of Westlaw headnotes to train its AI legal research tool was not fair use. The panel affirmed the district court’s partial summary judgment for Thomson Reuters on both questions certified for appeal: the headnotes are original enough to be protected by copyright, and ROSS’s copying was not excused as fair use.

The decision will be cited in AI copyright disputes for years. It is also narrower than many headlines suggest. ROSS’s system did not generate text; it retrieved passages from court opinions. The court itself described the dispute as “no more than an ordinary copyright case,” and it took care to distinguish the generative AI cases now shaping the field.

Key takeaways

The ruling does not hold that AI training is infringement, nor that it is fair use. It holds that copying a competitor’s protected editorial work to build a substitute product is not made fair by the fact that the copies were used to train an AI, particularly when the underlying public-domain material was freely available.

The court also found harm to a developing market for licensing Westlaw headnotes as AI training data. That finding is specific to this case, but it shows how courts may treat training-data licensing as those markets mature.

For readers following Japan, the reasoning has a close parallel in the proviso to Article 30-4 of the Japanese Copyright Act, which excludes uses that would unreasonably prejudice the interests of the copyright owner.

The dispute: headnotes, not court opinions

Judicial opinions are in the public domain in the United States. Westlaw headnotes are not. They are short summaries, written by Thomson Reuters editors, that state a point of law from an opinion clearly enough to stand on their own. Editors decide which points of law matter and how to express them, working within house guidelines that call for concise, self-contained statements.

ROSS, founded by three University of Toronto computer science students, set out to build a tool that would answer plain-language legal questions with relevant passages from court opinions. Before building its training data, ROSS had asked Thomson Reuters for a license, which Thomson Reuters declined to grant to a competitor. ROSS then engaged the legal services company LegalEase, which produced roughly 25,000 training memos. Each memo posed a legal question and paired it with passages from opinions ranked by relevance. Memo drafters built those questions from Westlaw headnotes because, according to the record, the headnotes offered an easy way to frame them.

ROSS made no secret of its ambitions. It advertised its platform in direct comparison with Westlaw, at prices in line with Westlaw’s, and some law firms switched.

Two figures in the case are easy to confuse. The district court found, as a matter of law, that 2,243 specific headnotes had been copied and infringed; those are the works at issue on appeal. The figure of roughly 25,000 refers to the headnotes the court described as copied in full into the training memos, and it bears on the court’s analysis of how much ROSS took.

The court held that all 2,243 headnotes clear copyright’s low bar for originality. Deciding which points of law to include and how to phrase them is a creative judgment, however modest. The court noted an amicus brief from RELX explaining that Lexis, a Westlaw competitor, uses its own process and produces different headnotes for the same opinions. When two editorial teams working from the same source produce different summaries, those summaries reflect choices, and choices are what copyright protects.

Primary source: U.S. Court of Appeals for the Third Circuit, Thomson Reuters Enterprise Centre GmbH v. ROSS Intelligence Inc., No. 25-2153

How the court applied the four fair use factors

Section 107 of the Copyright Act directs courts to weigh four factors. The panel found that only the second favored ROSS, and only slightly.

Factor	What the court found	Weighs
1. Purpose and character of the use	ROSS’s use was commercial and shared the same ultimate purpose as Westlaw: helping users find relevant legal material. The intermediate step of training an AI arguably made a slight difference, leaving the use “minimally transformative, at best.” Copying was a matter of convenience, not necessity.	Against ROSS
2. Nature of the copyrighted work	The headnotes were published before ROSS used them and are largely factual, since they must accurately describe the law.	Slightly for ROSS
3. Amount and substantiality	ROSS copied the full text of each headnote, and each headnote is a work in its own right. The court rejected ROSS’s argument that it had taken only about 0.08% of Thomson Reuters’s roughly 28 million headnotes. With the opinions freely available, ROSS took more than it needed.	Against ROSS
4. Effect on the market	ROSS diminished the headnotes’ value as a draw for Westlaw subscriptions and built a substitute in the same legal research market. The court also found harm to a rapidly developing market for licensing headnotes as AI training data, noting that Thomson Reuters already used its headnotes to train its own AI search products. ROSS’s claimed public benefits, including wider access to the law and national security, were unsupported by evidence.	Against ROSS

Two threads run through the analysis. The first is competition: ROSS used Thomson Reuters’s work to build a substitute for Thomson Reuters’s product, and the training step in between did not change that. The second is necessity. The court distinguished Google v. Oracle, Sega v. Accolade, and Sony v. Connectix, where copying was needed to reach unprotected functional elements of software. Here, the unprotected material, the opinions themselves, was freely available. As the court put it, “Unlike necessity, ease is not a justification for copying.”

The court also rejected ROSS’s reliance on Authors Guild v. Google. Google Books helped readers find books they might then buy; ROSS, by its own account, aimed to replace Westlaw.

What the ruling does not decide

The opinion is explicit about its limits. ROSS’s AI could not create new expression; it could only return passages from existing opinions. In a footnote, the court addressed the Justice Department’s statement of interest in In re OpenAI, the consolidated copyright litigation in the Southern District of New York, which relied on Bartz v. Anthropic to argue that training a large language model is transformative. The court said those concerns do not apply here, because ROSS’s system cannot generate original expression and was trained to serve as a commercial substitute for Westlaw. Bartz itself has since been resolved through a $1.5 billion class settlement, which received final approval in July 2026.

The facts also set the case apart. The copying took place before the current wave of generative AI, and ROSS has since shut down. The record also contains evidence of bad faith: ROSS attempted to access Westlaw using an investor law firm’s credentials after being told the terms of service prohibited it, and employees sought access while concealing that they worked for a competitor. The court noted that, to the extent good faith remains relevant to fair use, this evidence weighs against ROSS.

Procedurally, this was an interlocutory appeal under 28 U.S.C. § 1292(b). The Third Circuit affirmed the district court’s partial summary judgment on the two certified questions. That is a significant win for Thomson Reuters, but it is not the end of the litigation.

Editor’s view: the court protected the editorial layer

Read as a contest between an AI company and a rights holder, this case looks like a win for publishers. Read from an editor’s desk, it says something more precise. What the court protected was not the information, which was free, but the judgment of the people who selected, condensed, and framed it.

That distinction maps directly onto AI training data. Raw material and a curated dataset are not the same thing. A dataset that someone has selected, labeled, ranked, or annotated carries that person’s choices, and this ruling treats those choices as worth protecting, at least where they meet copyright’s threshold of originality.

The shortcut problem

ROSS could have drafted its training questions from the opinions themselves. It used the headnotes because that was faster. Creators who build their own fine-tuning or LoRA datasets face the same temptation. Third-party captions, tag sets, curated collections, prompt libraries, and preset packs are often more convenient than working from source material, and they are exactly the kind of editorial layer this case is about.

We think good data practice now means asking two questions instead of one: where did this material come from, and who organized it, on what terms? Whether a particular caption set or classification scheme is protected will depend on the facts and on the jurisdiction. The point is not that every annotation is copyrightable. It is that convenience, on its own, is not a defense.

Licensing markets reinforce themselves

At oral argument, ROSS warned that letting Thomson Reuters claim a lost licensing market would be circular: any copyright owner could say it would have licensed the very use being challenged. The court relied instead on evidence that a market for licensing headnotes as AI training text was developing quickly and that Thomson Reuters was already using its headnotes to train its own AI products. It added that Thomson Reuters’s decision not to license headnotes to others did not show that no such market existed.

In our view, this dynamic will compound. Every training-data license that gets signed makes the next market-harm argument easier to prove. Rights holders with the scale to negotiate those deals gain leverage, while individual creators who never state their terms risk being left outside a market that others are defining. Publishing terms does not create rights the law does not already provide. It does make clear which uses are welcome, which require permission, and which should be paid for.

A view from Japan: Article 30-4 and its proviso

Japan has no general fair use doctrine. Instead, Article 30-4 of the Copyright Act permits the use of works for purposes that do not involve enjoying the thoughts or sentiments they express, including data analysis such as AI training. For this reason, Japan is often described abroad as one of the most permissive jurisdictions for AI training.

That description is incomplete. Article 30-4 does not apply where a use would unreasonably prejudice the interests of the copyright owner. The Agency for Cultural Affairs’ General Understanding on AI and Copyright in Japan (2024) gives, as an example, reproducing a database work for data analysis such as AI training where licenses for that kind of analysis are available in the market. The guidance also discusses cases in which technical measures against AI training, together with other facts, indicate that such a database is likely to be offered for sale.

The legal tests differ, and a U.S. appellate decision carries no authority in Japan. Yet the two systems converge on the same pressure point. When data has been organized for analysis and a licensing market for it exists or is emerging, unlicensed copying is where the protection for AI training is weakest. For global companies training models on Japanese content, and for Japanese creators whose work is collected abroad, Article 30-4 should not be read as blanket permission.

Reference: Agency for Cultural Affairs, Government of Japan: Copyright

Conclusion: “AI” does not decide the case

The Third Circuit did not decide the future of generative AI training. It decided that a company may not copy a competitor’s editorial work to build a substitute product and then point to machine learning as the reason the copying was fair. The questions that decided this case will decide others: what was copied, who shaped it, why it was used, which market the result competes in, and whether a licensing market was bypassed.

Editor’s note

The source was free; the editing was not. That, in one line, is what this ruling says. The debate over AI training data is shifting from what a model learned to whose judgment it learned from. Creators who understand that shift will be better placed both to protect their work and to license it.

Sato Seiichi, Editor-in-Chief, AI Creators

Key references

en.ai-creators.tech

Generative AI Creative Services for Business

https://en.ai-creators.tech/enterprise/

AI Creators supports companies with generative AI creative production, AI creator matching, corporate training, strategic advisory, PR, and partnerships.

The post Thomson Reuters v. ROSS: Appeals Court Rejects AI Fair Use first appeared on AI Creators.

GMI & MiniMax AI Art Gallery: Submit Your AI Film by Sept 30

aratama 璞 | Editor-in-Chief, AI Creators — Sat, 19 Sep 2026 19:00:15 +0000

Experience AI-generated films and music on a big screen and connect with fellow creators. The GMI & MiniMax AI Art Gallery takes place at Shibuya Hikarie in Tokyo on Thursday, October 8, 2026.

The evening will feature screenings, networking, and a talk by special guest Kiyoshi Shin. Creators are also invited to submit videos made using GMI Cloud or MiniMax models, with entries due Wednesday, September 30.

Admission is free, with advance registration and organizer approval required. You can apply to attend whether you are submitting a film, interested in discovering AI-generated work, or looking to connect with creators and engineers working with generative AI.

At a Glance: Attend or Submit Your Work

Attend the event: Thursday, October 8, from 6:00 to 8:00 p.m. JST. Register through the official Luma page and wait for approval from the organizer.
Submit a film: Entries close Wednesday, September 30. Submissions must be 1–3 minutes long and include audio or music.
How to enter: Post your work on X, then submit the post URL through the event registration form.

Visit the official event page to register or submit your work.
(Source: GMI & MiniMax AI Art Gallery | Luma)

AI Screenings and Networking at Shibuya Hikarie on October 8

The GMI & MiniMax AI Art Gallery brings AI-generated video, music, and art into a shared, in-person setting.

Enjoy screenings over food and drinks while meeting the creators behind the work, AI engineers, and professionals working across video, music, and design. Selected works are scheduled to be shown on a 100-inch screen.

The event is aimed at people seeking opportunities to showcase their work, discover other creators’ approaches, or explore collaborations involving generative AI.

Event Details

Item	Details
Event	GMI & MiniMax AI Art Gallery
Date and Time	Thursday, October 8, 2026, 6:00–8:00 p.m. JST
Check-in Opens	5:30 p.m. JST
Venue	Shibuya Hikarie, Tokyo. The organizer will provide venue details after registration.
Admission	Free
Registration	Advance registration through the official Luma page is required, subject to organizer approval. (Source: GMI & MiniMax AI Art Gallery \| Luma)
What to Bring	One business card
Organizer	GMI Cloud Japan
Sponsors	NVIDIA Japan and MiniMax

The event welcomes creators and engineers using generative AI in video, music, and design, as well as marketing, PR, advertising, and production professionals seeking to collaborate with AI artists and creators. Anyone involved in—or interested in—AI-powered creativity is invited to apply to attend.
(Source: GMI & MiniMax AI Art Gallery | Luma)

Kiyoshi Shin on Bringing Generative AI into Creative Production

Alongside opening remarks from GMI Cloud and MiniMax, the event will feature Kiyoshi Shin as a special guest speaker.
(Source: Kiyoshi Shin (@kiyoshi_shin) | X)

His session will explore the latest generative AI technologies, including MiniMax H3, and practical ways to use them in creative production, drawing on his experience as a creator.

According to the organizer’s speaker biography, Shin is affiliated with Creative Tech Lab. at Baleen Studio and is a professor at Digital Hollywood University Graduate School. Following his work as a games journalist, he moved into game development, leading development of the VR title Swords of Gargantua.

He also worked on AI-assisted character design and 3D prop creation for the crafting and survival game Exelio. For those exploring how to integrate generative AI into production workflows, the session offers an opportunity to hear insights grounded in hands-on experience.
(Source: GMI & MiniMax AI Art Gallery | Luma)

Submit Your AI Film by September 30 for a Chance to Screen at the Event

The organizers are accepting video submissions with audio or music, created using GMI Cloud or MiniMax models.

The MiniMax and GMI Cloud teams will select works to screen at the event. Selected works will be revealed through the screenings themselves on the day.

Submission Requirements

Item	Requirement
Tools or Models	Work created using GMI Cloud or MiniMax models
Content	Video with audio or music
Aspect Ratio	16:9
Duration	1–3 minutes
File Format	MP4
Resolution	4K (3840 × 2160 px) recommended; Full HD (1920 × 1080 px) minimum.
Submission Deadline	Wednesday, September 30, 2026 (Source: GMI & MiniMax AI Art Gallery \| Luma)

Works are scheduled to be shown on a 100-inch screen. Before submitting, check your video’s resolution and aspect ratio, and make sure it includes audio or music.

How to Enter: Add Your Credits, Post on X, and Submit the Link

Add your creator details to the end of your video.
Include your name or creator handle, X account, or other identifying information at the end of the work.
Post the work on your own X account.
Mention both GMI Cloud and MiniMax in the text of your post.
Submit the X post URL through the event registration form.
When registering on the official Luma page, enter the link to your work’s X post in the registration form.

A Note from the AI Creators Editorial Team

Prepare your video with creator credits and have your X post URL ready before starting event registration. Posting on X alone does not complete your entry—make sure you submit the link through the registration form.

The organizers state that people who submit work created using GMI Cloud or MiniMax will receive priority consideration for invitations. However, submitting a work does not automatically confirm your attendance; organizer approval is still required.
(Source: GMI & MiniMax AI Art Gallery | Luma)

What to Check Before Submitting

Submissions must be work you created yourself or with your team. Before entering, ensure you have the necessary rights and permissions for all video, images, music, characters, and other material used in the work.

By submitting, you agree to have your work screened at the event.
The organizers will select the screening lineup. Not all submissions will be shown.
Works that infringe third-party rights, violate public order or accepted standards of decency, or are otherwise deemed unsuitable for the event by the organizers will not be screened.
Selected works will be announced through their screening at the event.

According to the organizers, submitted work files will not be used for purposes other than screening at this event. However, posts that entrants have publicly shared on X may be reposted or shared by the official GMI Cloud and MiniMax social media accounts. Please review these terms before submitting.
(Source: GMI & MiniMax AI Art Gallery | Luma)

MiniMax Tote Bags for Attendees and Merchandise for Award Recipients

MiniMax-branded tote bags are planned as attendee gifts, available on a first-come, first-served basis. Quantities are limited, and distribution will end when supplies run out.

The organizers have also announced MiniMax merchandise for creators whose submissions are chosen for awards, including:

T-shirts
Caps
Little M
Stickers

Award recipients will be announced and gifts presented at the venue on the day of the event. These awards are not gifts for every entrant.

Free Admission with Advance Approval: Confirm Your Place Before Attending

Attendance requires advance registration through the official Luma page and approval from the organizer. Submitting a registration request does not confirm your place. Only approved guests may attend, so please check the organizer’s confirmation before coming to the venue.

Capacity is limited, and the organizers state that a lottery may be held if applications exceed available places. After registering, check your attendance confirmation and venue details, and bring one business card on the day.

The organizers also ask attendees to note the following:

Speakers, session content, and event times may change without advance notice.
The organizer and sponsors reserve the right to decline attendance, including applications from competing agencies or companies handling competing products.
Photography and video recording will take place at the venue.
Photos and footage may be used on the organizer’s website and social media, and in future event promotions and publicity.

Please review the attendance requirements and notices on the official event page before registering.
(Source: GMI & MiniMax AI Art Gallery | Luma)

About GMI Cloud

GMI Cloud is a cloud infrastructure company supporting AI applications. According to the organizer, its AI inference platform provides browser-based access to generative AI models including MiniMax, Flux, Seedream, and Kling, allowing users to compare and evaluate outputs across text, images, video, and audio.
(Source: GMI Cloud Official Website | GMI Cloud)

The event offers an opportunity to experience AI-generated work in person and talk directly with creators and engineers about their approaches. Whether you want to showcase your own work or discover what others are making, visit the official page to apply to attend.

Event: Thursday, October 8. Submission Deadline: Wednesday, September 30.

Admission is free, with advance registration and approval required. To submit a work, review the entry requirements, post your video on X, and send the post URL through the event registration form.

Visit the official page for full details and registration.
(Source: GMI & MiniMax AI Art Gallery | Luma)

This article was edited and produced by AI Creators using materials supplied by the organizer and information on the official event page. Please refer to the official event page for the latest event details and submission requirements.

en.ai-creators.tech

AI Media Distribution, PR & Creator Campaigns

https://en.ai-creators.tech/enterprise/advertising/

Media distribution, sponsored content, bilingual PR, and AI creator campaigns for AI companies seeking audiences in Japan and worldwide.

The post GMI & MiniMax AI Art Gallery: Submit Your AI Film by Sept 30 first appeared on AI Creators.

CapCut World 2026 Singapore: Skills, Hackathon & Creator Awards

aratama 璞 | Editor-in-Chief, AI Creators — Thu, 10 Sep 2026 08:01:50 +0000

Disclosure: The author attended CapCut World 2026 Singapore as an invited AI creator and also took part in the CapCut AI Creators Hackathon. Travel and accommodation were provided by CapCut. This report was independently structured and written by the AI Creators editorial team.

CapCut’s first global community event series opened in Singapore in September 2026, and it did not look much like a conventional software launch.

There was a keynote, and there were new tools. But the event extended well beyond product announcements: a free public experience at Marina Bay Sands, an invitation-only program at Capella Singapore, AI Festival screenings, creator sessions, a red carpet, an awards gala, an after-party, and a hackathon in which teams of creators were given roughly 50 minutes to turn an idea into a finished film.

According to CapCut, more than 300 creators, industry professionals, and CapCut team members gathered in Singapore.
(Sources: Celebrating Our Community’s Growth at CapCut World in Singapore | CapCut Newsroom / CapCut World 2026)

Reading the event simply as “a big global CapCut gathering” misses what was actually on display. Three developments mattered more than any single feature announcement.

First, Skills — a mechanism for saving and sharing not only finished work, but the styles, workflows, and techniques behind it. Second, a creative pipeline extending from AI Lab through Infinite Canvas, generation, editing, and Edit Pilot, presented as a more continuous workflow rather than a collection of disconnected tools. Third, three distinct systems of recognition operating within the same event: Creator Awards, the CRE[AI]TE AI Festival, and the on-site hackathon.

What Singapore showed was not simply a video editor adding AI features. It was an attempt to connect creative output, production process, execution environment, community, and recognition into a broader creator economy.

Item	Details
Event	CapCut World 2026 — CapCut’s first global community event series
Location	Singapore — first stop in the series
Public program	Marina Bay Sands / September 4–6, 2026 / free entry AI Studio, Design Studio, Edit Studio
Invitation-only main program	Capella Singapore / September 5, 2026
Key programs	Keynote, Breakout Sessions, CapCut AI Festival screenings, Creator Workshop, Creator Awards, After Party
Attendance	300+ creators, industry professionals, and CapCut team members combined
Organizer	CapCut (ByteDance)
Next stops	Los Angeles (October 2026) / London (November 2026)

*The “300+” figure includes creators, industry professionals, and CapCut team members combined; it does not refer to creators alone.
(Sources: CapCut Newsroom / CapCut World Official Site)

What Was CapCut World 2026 Singapore Really About?

この投稿をInstagramで見る

CapCut(@capcutapp)がシェアした投稿

Most software-company events put new features and roadmaps at the center. CapCut World put the creators using the product there instead, with product experiences built around a broader community program.

Three Cities, One Global Event Series

CapCut World is being staged across Singapore, Los Angeles, and London. Product experiences, creator workshops, screenings, live showcases, and awards all sit within the same broader format.

The questions raised by the programming went beyond features. Who uses these tools? How do they create? How does a finished work — or a working method — move from one creator to another? How do creators learn from one another? And what kinds of work and activity ultimately receive recognition?

The creative activity surrounding the product was itself part of what CapCut World was designed to showcase.
(Source: CapCut World 2026 | CapCut)

From Singapore to Los Angeles and London

Singapore was the first stop, with Los Angeles scheduled for October 2026 and London for November. In Singapore, I met creators from across Asia, including Vietnam, Indonesia, the Philippines, Korea, and Japan. The series then moves to the United States and the United Kingdom.

Rather than viewing this simply as the same regional event repeated three times, it can also be read as an attempt to connect distinct creator communities through a shared global format. How much of CapCut World remains consistent across Singapore, Los Angeles, and London — and where local creative culture begins to reshape it — will be one of the more interesting things to watch.
(Source: CapCut World 2026 | CapCut)

Two Venues, Two Audiences

CapCut World 2026 — DAY 3 🇸🇬
旅の最後は、Marina Bay SandsのCapCut World POP-UPへ。

AI Studio、Design Studio、Edit Studioなど、一般公開されたさまざまなクリエイティブ体験を巡りながら、Singaporeでの最後の一日を過ごしました。

Capella… pic.twitter.com/FDSoXPbAYI

— aratama 璞 (@aratamadao) September 6, 2026

One of the defining structural choices in Singapore was that the event did not take place at a single venue.

At Marina Bay Sands Waterfront Event Plaza, from September 4–6, AI Studio, Design Studio, and Edit Studio were presented as part of a free public experience. At Capella Singapore on September 5, the program was invitation-only, with the keynote, breakout sessions, CapCut AI Festival screenings, a creator workshop, awards, and an after-party.

A public creative experience and an invitation-only creator and industry program, operating under the same CapCut World umbrella.

The significance was not simply that both existed. The public program centered on AI, design, and editing, while the Capella program brought creators and industry professionals into deeper sessions, screenings, workshops, and awards. Professional production and everyday creativity were therefore placed within the same overall event architecture.
(Sources: Three days. Two venues. | CapCut on X / CapCut World Singapore | CapCut Japan)

Why Product, Screenings, and Community Belonged Together

These could easily have been separate events. A product launch can end once the new features have been explained. A festival can focus on screenings and awards. A community event can center on networking.

CapCut World instead placed discovery, experimentation, production, collaboration, screenings, recognition, and public creative experiences inside the same event ecosystem.

Use the product. Create. Share. Learn. Receive recognition. Then create again. Seen alongside Skills and the three recognition systems discussed later in this report, that cycle becomes one of the most useful ways to understand the event.

Beyond 660 Million Users: CapCut’s Template Economy

CapCut says it now serves more than 660 million users across more than 170 markets and 30 languages. That is an extraordinary distribution figure, but user count alone does not explain the structure of the platform.

The more revealing numbers are on the supply side. According to CapCut’s published figures, the platform has more than six million template creators, with over 400,000 new templates published every day.
(Source: CapCut / ByteDance Press Release)

That is different from how a conventional editor operates. Normally, a user creates something in editing software and publishes the completed file elsewhere. On CapCut, a structure created by one person can become the starting point for another person’s creation inside the same platform.

One creator’s output becomes another creator’s infrastructure.

At more than 400,000 new templates a day, that loop is large enough to function as a market in its own right. It also broadens what the word “creator” can mean: not only someone who produces finished work, but someone who supplies reusable structures, assets, and formats that enable other people to create.

CapCut already had an economy in which finished structures could circulate. What CapCut World made easier to see was a second layer that can increasingly circulate: the process behind the work.

CapCut Skills: When the Creative Process Becomes Reusable

Of the ideas CapCut brought together in Singapore, Skills is the most consequential. It shifts the unit of reuse from the output toward the method.

What Templates Share

A template lets someone reuse a finished structure: an arrangement, effects, transitions, and other elements close to the completed result. What it does not necessarily carry is the reasoning behind that result — the sequence of decisions, the workflow, or the method used to maintain a consistent look.

What Skills Shares

The Skills feature allows creators to save, reuse, and share styles, workflows, and techniques. The unit shifts from what was made toward how it was made.

CapCut’s own example is filmmaker Vee Camallere and the AI series Featherweight, which has generated more than 25 million views. CapCut says the color-grading process was saved as a Skill, allowing the same look to be recalled across episodes.
(Source: Helping Community Members Iterate Together | CapCut Newsroom)

Templates	Skills
Reuse a finished structure	Reuse a production process
Output	Process
“What was made”	“How it was made”
Finished structures circulate	Methods and know-how become reusable

Anyone working to a deadline knows that a repeatable process can be more valuable than a one-off result. Color treatment. Character consistency. Brand style. Product-visual pipelines. Repeatable production routines for social content. If these can be stored and shared through Skills, experience that once lived inside one creator’s head or project files can become a reusable asset.

At that point, the question “Is CapCut becoming a Creator OS?” begins to feel less theoretical. It is still too early to make that claim outright. But a platform that increasingly stores not only what creators make, but also how they make it, is doing something meaningfully different from a conventional video editor.

Singapore Was an Integration Event, Not Simply a Launch Event

It is important to be precise about what appeared when. Skills and the surrounding infrastructure did not suddenly emerge on the day of CapCut World Singapore.

Date	Key Development
July 8, 2026	Seedream 5.0 Pro — image model with region-based editing, layer separation, multilingual text generation, and other capabilities
July 31, 2026	Dreamina Seedance 2.5 / Dola Seedream 5.0 Pro — announced alongside C2PA Content Credentials, invisible watermarking, and visible AI labels
August 10, 2026	Skills rolled out for Design Studio, positioning repeatable workflows as reusable creative capabilities
September 4–6, 2026	CapCut World 2026 Singapore

These foundational releases arrived over roughly five weeks, followed by CapCut World less than four weeks after the Skills rollout. The generative models, provenance measures, and process-reuse mechanism had therefore appeared separately before creators arrived in Singapore.

CapCut World was less a stage for unveiling every capability for the first time than a stage for assembly — a place where previously separate pieces were shown as parts of one creative workflow.

That changes the more useful question from “What was announced?” to “How do the parts connect?”
(Sources: Seedream 5.0 Pro | CapCut / Dreamina Seedance 2.5 and Dola Seedream 5.0 Pro | CapCut Newsroom / CapCut Design Studio Levels Up | GlobeNewswire)

From AI Lab to Edit Pilot: An Integrated Creative Workflow

Say hello to the New Way of Content Creation with CapCut! 👏

Built on the video editor you already know, CapCut is now an all-in-one AI-powered creative platform, with video and image editing, AI generation, filmmaking and design together in one place, on mobile, desktop, web… pic.twitter.com/Vork6j69Il

— CapCut (@capcutapp) September 9, 2026

The more consequential part of CapCut’s AI strategy may not be any single model. It is the effort to carry creative context from the first idea through to the final edit with as little interruption as possible.

The Creative Pipeline

Idea → AI Lab → Infinite Canvas → story, character, and shot development → AI generation → multitrack editing → Edit Pilot → final output

AI Lab supports ideation in natural language. Infinite Canvas is positioned as a space where stories, characters, shots, and scenes can be developed. That context and those assets can then move toward the editing environment, where Edit Pilot assists the production process. Dreamina Seedance supports video generation, Dola Seedream supports image generation, and Seed Audio covers sound generation.

CapCut says it now offers more than 85 AI tools, that more than six billion images, video clips, effects, and other outputs have been generated in 2026, and that more than seven million creators use AI features each day.
(Source: Empowering People Throughout the Creative Process | CapCut Newsroom)

The number 85 is not the most interesting part. Eighty-five disconnected tools would still leave creators exporting assets, re-entering prompts, and rebuilding context between environments. The more important question is how much of that friction can be removed.

Generative AI is moving deeper into the editing environment, while ideation and story development are also moving closer to the same workflow. That is more significant than simply adding another AI feature to an editor.

AI Provenance Is Becoming Part of the Product

The deeper generation is integrated into a platform, the more important provenance becomes. What was AI-generated? And how can that origin be communicated?

CapCut has outlined transparency measures including invisible watermarking, C2PA Content Credentials, and visible AI labels. These measures were introduced alongside its model announcements rather than only as a separate policy layer.

Combining increasingly capable generation with reliable provenance remains an industry-wide challenge. For professional AI creators, provenance is becoming a consideration alongside output quality itself.
(Source: Giving Creators More Control with Dreamina Seedance 2.5 and Dola Seedream 5.0 Pro | CapCut Newsroom)

CapCut AI Creators Hackathon Case Study: ROOM 0 in 50 Minutes

Title	ROOM 0
Created at	CapCut AI Creators Hackathon / CapCut World 2026 Singapore
Runtime	1 min 30 sec
Production time	Approx. 50 minutes
Evaluation	Participant vote after all works were screened
Recognition	Best Main Character Design — Hackathon
Team	aratama 璞 / Bird / Okada / SHAINA
Special thanks	amity_sensei
Synopsis	Four creators race through a series of impossible games against a deadly countdown.

The hackathon became one of the most revealing parts of the event because it put the creative workflow under real production pressure. A workflow can appear seamless on a keynote stage; a 50-minute deadline makes its practical bottlenecks much easier to see.

The Production Process — Nine Steps

Before the Hackathon: From Team Announcement to the Night Before

Teams announced several weeks ahead
We learned who we would be working with. None of us had met in person. We had exchanged messages on X, but they were largely logistical rather than creative.
A story meeting at a Singapore street-food stall
Three of us were able to meet after arriving and went to a street-food stall, where we talked for several hours. That was where the central idea emerged: take our own situation — creators who had to finish something before the clock ran out — and turn it into the film itself. A death game.

Event Day: Before the Clock Started

Bringing our own material
The full team effectively introduced themselves for the first time that day. Everyone brought original character sheets, background assets, and other material, including our original CapCut Man character. We were not inventing every character from scratch during the hackathon. Each creator brought characters they had already been developing independently.

The Hackathon: 50 Minutes

Start
We had roughly 50 minutes. Teams were seated close together, with staff, participants, and other people around us able to see the work unfolding. That created a kind of live pressure very different from a normal production environment.
Shot design
We worked backward from a 90-second target: how many scenes were required, how long each could run, which characters would appear, and what each prompt needed to accomplish. The governing question became less “What can we make?” and more “What can we actually finish?”
Parallel generation
We divided the scenes and generated them simultaneously. Usable shots went directly into the edit; weaker ones were regenerated. Waiting for every shot to reach an ideal state would have meant missing the deadline.
Structuring the film around one game per scene also made selection easier because each sequence could be judged independently.
Music
The music was also created on-site. A 90-second film needed an audio arc as well as a visual one, so sound had to develop alongside generation rather than afterward.
Editing while still generating
Finished shots went straight onto the timeline as they arrived. We did not wait until generation was complete before starting the edit. Under this kind of deadline, that parallelization was critical.
Submitting at the deadline
The title and English one-line introduction still had to be written, followed by the upload and submission itself. A concept is not a finished film. Generated footage is not a finished film. The work is finished when it is delivered.

The films were screened immediately afterward and put to a participant vote. We had imagined more games and additional ideas for the ending than we ultimately used. In retrospect, what we were forced to cut shaped ROOM 0 as much as what remained.

Four creators trapped in a deadly countdown must complete a series of impossible games—only to discover that the final test was never about survival, but about creating the next moment together.

Why ROOM 0 Received Best Main Character Design

ROOM 0 received Best Main Character Design within the hackathon. Given the way the film was made, the recognition was closely connected to the production method itself.

Four Separate Creative Worlds, One Film

No single creator designed the full cast from scratch in one unified style. Each character came from a different creator’s existing practice. Different visual languages. Different settings. Different creative sensibilities. We used the character sheets almost as a shared production specification and brought those characters together inside a single 90-second film.

Character consistency remains a major challenge in generative video. Producing one compelling character is not the same problem as preserving that character as recognizably the same person across multiple shots.

That is why character sheets matter. A front view alone is rarely enough. Angles, expressions, clothing, body proportions, and identifying details all help establish a stable reference. The objective is not to redesign the character with every generation, but to define an identity standard first and generate against it.

ROOM 0 is therefore better understood not as a film that invented characters with AI, but as a film that integrated existing character IP through an AI workflow. As generating a character becomes easier, questions of authorship, ownership, identity, and consistency become more important rather than less.

Creator Awards: How CapCut Recognizes Creator Activity

CapCut World 2026 — DAY 2.2 🇸🇬
19:00からは、Capella Singaporeで Red Carpet, Awards Gala & After Partyへ。

世界中から集まったクリエイターたちが一堂に会し、Red CarpetからCreator Awards、そしてAfter Partyまで続く特別な夜でした。

CapCut… pic.twitter.com/svJYunCYXj

— aratama 璞 (@aratamadao) September 6, 2026

The Creator Awards took place at Capella Singapore on the evening of September 5. What they recognized was different from what the hackathon evaluated.

The Creator Awards spanned eight categories: Asset Member Favorite, Boundless Explorer, Creative Impact, AI Visionary, Creator of the Year, Top Partner, Community Star, and Trendsetter.

The category names themselves suggest that this was not simply a competition between individual finished works. Categories such as Top Partner, Community Star, and Trendsetter point toward ongoing platform participation, community contribution, and influence over time.

Vietnam’s state broadcaster VTV reported that 20 Vietnamese creators were recognized across the eight categories. Kim Hoa was named Creator of the Year, while BatToS, BMSound, Phong AI, Shjn, and Bé Cừu were among the AI Visionary recipients.
(Source: Việt Nam dẫn đầu CapCut World Award tại Singapore | VTV)

Why Were 20 Honorees From Vietnam?

This should not be reduced to a simplistic comparison in which Vietnamese creators are somehow “better” than creators elsewhere. The Creator Awards were not designed around finished-work quality alone.

A more useful lens is the depth of the regional creator economy. How many creators supply templates? How many maintain an ongoing relationship with the platform? How connected are creators to one another as a community? And how effectively does the platform discover, support, and recognize creators in each region?

Award counts are not a clean proxy for the overall health of a creative scene. But 20 honorees is still a meaningful reminder that a creator’s standing is not determined by production skill alone. Creation, publishing, community participation, distribution, and sustained activity increasingly operate as separate but connected layers.

CRE[AI]TE: Evaluating AI Work as Creative Work

The canvas is ready. What will you CRE[AI]TE?

We are excited to launch CapCut’s 1st AI Festival, a global celebration of original work and the imagination AI makes possible.

Submissions are now open through August 10, 2026.

From Films and Series to Creative and Commercial,… pic.twitter.com/SHS8dl9Kbz

— CapCut (@capcutapp) June 10, 2026

While Creator Awards focus on creators and their broader activity, CRE[AI]TE evaluates finished creative work. Launched in 2026 as CapCut’s first AI Festival, it is structured as an international competition.
(Source: CapCut CRE[AI]TE 2026)

Four Categories

Film — a complete story with characters, cinematic craft, and a clear point of view
Series — work that builds a shared universe across multiple episodes
Creative — visually driven, experimental work
Commercial — spec ads and brand films for fictional or self-created brands

What is notable is that there is no separate “AI Movie” or “AI-Generated Video” category. The structure maps closely to established creative-industry formats: Film, Series, Creative, and Commercial.

AI is not the genre. What you make with it is the genre.

Cash Awards and Selection Structure

CRE[AI]TE recognizes 30 works in total: 10 Cash Awards and 20 Spotlight Selections.

Award	Recipients	Amount
Grand Prix	One work across all categories	US$70,000
Category Winner	One per category × 4	US$20,000 each
Category Honoree	One per category × 4	US$10,000 each
Student Vision Award	One student	US$10,000

The itemized amounts in the current Program Terms add up to US$200,000. The official Terms do not state a headline total as a standalone figure, so this report treats US$200,000 strictly as the sum of the published award breakdown.

Selected creators or teams are also listed as receiving 50,000 CapCut Credits and one year of CapCut Pro, subject to the applicable conditions. Because program terms may change, creators should refer to the latest official rules.
(Source: Program Terms | CapCut CRE[AI]TE 2026)

The Judging Criteria Matter More Than the Prize Money

The most revealing part of CRE[AI]TE is not the size of the prizes. It is what the work is judged on.

The Program Terms list Human Vision and Intent, Story or Concept, Craft with AI, Originality and Imagination, Taste, Sound, and Overall Execution as judging criteria.

In other words, the center of gravity is not “How much AI did you use?” It is story. Intent. Craft. Originality. Taste. Sound. Execution.

CapCut appears to be moving the evaluation of AI video away from technical demonstration and toward creative work. More importantly, that direction is expressed through category design, judging criteria, and awards — not only through marketing language.

As AI filmmaking matures, “Can you generate it?” increasingly becomes the baseline. The questions above it are familiar ones: Why this shot? Why this character? Is there a reason to keep watching? Do the picture, sound, and edit function as one coherent work?
(Source: Judging Panel Criteria | CapCut CRE[AI]TE 2026)

What Can Be Confirmed About Singapore

CapCut World’s program confirms that CapCut AI Festival screenings took place at Capella Singapore on September 5, and VTV also reported that CRE[AI]TE 2026 works were screened there.

What cannot currently be confirmed from public sources is whether the formal CRE[AI]TE winner announcement took place on that stage. This report therefore treats the screening of CRE[AI]TE / CapCut AI Festival work in Singapore as the confirmed fact and does not go beyond it.
(Source: VTV)

Hackathon, Creator Awards, and CRE[AI]TE: Three Recognition Systems

この投稿をInstagramで見る

CapCut(@capcutapp)がシェアした投稿

Put the three side by side and the distinction becomes much clearer.

	Creator Awards	CRE[AI]TE	Hackathon
Primary subject	Creator	Work	Team / live production
Time horizon	Sustained activity	Open competition	Short, time-constrained production
Decision process	Not disclosed in the public materials reviewed	Jury, using published criteria	Participant vote observed on-site
Evaluated on	Impact, community, trends, and related activity	Vision, story, craft	Execution, collaboration, result
Example	Creator of the Year	Grand Prix	Best Main Character Design
Primary value dimension	Community	Authorship	Process and collaboration

Calling all three simply “the awards” flattens what is most interesting about them.

One system recognizes the creator. One recognizes the work. One recognizes the ability to make something with other people, live, under pressure.

They also operate on different time horizons. A work can be evaluated as a single submission. A creator’s activity accumulates over months or years. Collaboration is tested in the room, in the moment. Different timescales require different forms of recognition.

The level of transparency is also different. CRE[AI]TE publishes its judging criteria. I observed the hackathon voting process directly on-site. The Creator Awards selection methodology was not detailed in the public materials reviewed for this article.

For clarity, ROOM 0’s Best Main Character Design was neither a Creator Award nor a CRE[AI]TE prize. It was an in-hackathon recognition. Keeping those systems separate is important to documenting CapCut World accurately.

CapCut’s Creator Community: When an Online Network Becomes Physical

The moment I remember most clearly from Singapore was not on the Capella stage. It was in the lobby of Hotel Ora, where many of the invited creators were staying — and before that, in the arrivals hall.

Creators kept appearing whose names I had previously known only from timelines. Their fields were completely different: AI film, templates, design, short-form video, fashion, visual art, marketing. People who would normally occupy separate algorithmic feeds were suddenly staying in the same hotel, riding the same shuttle, and sitting around the same table.

The same was true for the Japanese contingent. Everyone had rearranged something in their lives or work to make the trip to Singapore. Some left with results they were proud of; others may have left with frustration. Either way, after the formal program ended, many Japanese creators naturally gathered in the hotel lobby and continued talking late into the night.

“Creator community” is a term that gets used loosely. But ultimately, it comes down to people. Follower counts, online recognition, and connections are part of that picture. So is the experience of eating together in another country, creating side by side, and sharing the same setbacks and sense of achievement. That was where the value of community became tangible.

The Event Operation Was Part of the Product Experience

Watching CapCut World at close range, it would be a mistake to focus only on the stage program.

Multilingual support. International flights. Hotels. Airport reception. Dedicated shuttles. Hospitality desks. Badges, welcome kits, meals, transfers to Capella, workshops, awards, performances, the public Marina Bay Sands experience, and coordinated social campaigns around the event. There was an enormous operational layer behind the visible program.

Why Operations Matter as Product Experience

At a creative event, strong content can still be undermined by broken logistics. That becomes especially important when guests are arriving internationally. Flights, immigration, hotels, airport transfers, and on-site coordination may not appear “creative,” but failures in any of those areas can shape the entire event experience. Risk management is part of the design whether audiences see it or not.

For a creator community to operate globally, the operational layer also needs a degree of repeatability. If the creator experience changes dramatically from one city to another, it becomes difficult to sustain a coherent global community strategy.

CapCut World also continues beyond Singapore to Los Angeles and London. From a marketing perspective, the experience extends beyond campaign creative or the stage itself. The journey from a creator’s arrival through their return home can become part of the brand experience.

What Is CapCut Becoming? A Five-Layer Model

Taken together, the developments around CapCut World can be read as five connected layers. This is not an official CapCut framework or product term. It is an analytical model developed by the AI Creators editorial team based on what was shown in Singapore and on publicly available information.

Layer 1 — Output: Templates

One creator’s finished structure can be reused by another. More than six million template creators and over 400,000 new templates a day suggest that this layer already operates as a substantial economy.

Layer 2 — Process: Skills

The Skills feature moves reuse beyond the finished output toward the style, workflow, and technique behind it. If Templates circulate results, Skills turns methods into reusable assets.

Layer 3 — Execution

AI Lab, Infinite Canvas, Dreamina Seedance, Dola Seedream, Seed Audio, Editor, and Edit Pilot. Ideation, design, generation, revision, and editing are being brought closer together inside a more continuous production context.

Layer 4 — Community

Creators share Templates and use Skills to exchange methods and influence how other creators work. This layer includes distribution, but also learning. Once creators can absorb not only someone’s output but also the process behind it, the platform itself can begin to function as a learning environment. If the Skills feature scales, the significance of this layer will grow with it.

Layer 5 — Recognition

Creator Awards recognize creator activity. CRE[AI]TE recognizes finished work. The hackathon recognizes execution and collaboration under live constraints.

In a creator economy, recognition is not merely ceremonial. It creates visibility, visibility can create opportunity, and opportunity can change a creator’s role within a community. Create, share, receive recognition, create again. Recognition can be understood as the layer that closes the loop.

Why CapCut Can Connect These Five Layers

CapCut can bring these layers close together because it is not starting from zero. Skills and an integrated AI workflow are being added to an existing foundation: more than 660 million users, six million-plus template creators, over 400,000 daily templates, and more than 85 AI tools.

A strong AI model alone does not create a creator economy. Neither does a strong editor.

A sustainable loop emerges when tools, creators, reusable assets, community, and recognition reinforce one another.

CapCut already has a large reuse loop built around Templates. Extending that loop from output into process gives Skills more significance than the same feature might have on a platform without an existing creator economy.

And if creative context increasingly carries from AI Lab through to Editor, creators may have more reasons to keep their production process inside the platform. That can create network effects — but it also introduces a separate issue: platform dependency.

Can CapCut Be Called a Creator OS?

Once these five layers are visible, the question “Is CapCut simply a video editing app?” begins to feel somewhat outdated. Xiaoqi Fu, Global Head of CapCut, has described a trajectory that begins with making video creation more accessible and extends toward bringing editing, design, and newer creative tools into a more unified creative journey.
(Source: CapCut World Singapore | CapCut Newsroom)

Still, “Creator OS” is a large claim, and it would be premature to make it definitively. An operating system can imply broader conditions such as interoperability, independence from a single vendor, and a third-party developer ecosystem. What CapCut currently appears to be building is a deeply integrated creative journey inside its own services, which is different from a neutral infrastructure layer.

“Creator OS” stops being a loose analogy and becomes a claim that can be tested.

The Other Side: What It Means to Keep Your Creative Process on a Platform

An integrated workflow is genuinely convenient. It also increases what a professional creator needs to understand about where their work and process live.

CapCut’s Trust Center currently states: “All user-generated content you create on our platform belongs to you.” It also states that CapCut does not use a creator’s content or likeness for commercial purposes without consent. At the same time, it explains that sharing content on the platform requires limited permissions to enable functions such as downloading, reposting, sharing, and collaboration.
(Source: Legal Protections for Your Creative Works | CapCut Trust Center)

Neither extreme interpretation — “there is nothing to think about” or “you lose everything you upload” — is useful. A working professional instead needs a clear understanding of their own boundaries.

What goes to the cloud? Does client material belong there? Which assets are cleared for commercial use? How much of a character IP should live inside one platform? Can a workflow be moved somewhere else?

The Skills feature makes those questions more important. Backing up a finished JPEG or MP4 is relatively straightforward. But when value also lives in canvas context, AI interactions, a workflow saved as a Skill, or platform-specific editing behavior, portability becomes a different problem.

This is not unique to CapCut. It is a broader issue facing AI-native creative platforms.

Convenience and lock-in are often two sides of the same design.

For professional AI creators, the required skill set therefore extends beyond craft. It also includes judgment about how much IP, data, and working process to place inside someone else’s platform.

Seven Takeaways for AI Creators Entering the Global Market

Team Japan

It would be easy to end a report like this by saying that it was valuable to attend an international event. That would miss the more useful conclusion.

From a Japanese perspective, I did not leave Singapore thinking that Japan’s AI creators lack production ability. The more important challenge is how that ability connects to the world outside Japan. At the same time, many of the lessons from Singapore extend well beyond any single market.

Enter the global market directly
Publishing work internationally online is not the same as physically entering the same room. English does not need to be perfect for work and production experience to become a shared language. At the same time, conversations and collaborations emerge in person that would never happen through a timeline alone. A creator’s positioning and demand can also look different from one market to another, which makes objective self-assessment increasingly important.
Build original IP and clear differentiation
The ability to generate visually impressive work will become less scarce. The more important question is what you already own and what is recognizably yours. ROOM 0 benefited from each creator bringing an original character into the project. Creators also need to ask what is distinctive about their perspective in a global context — and what remains distinctive at the individual level.
Treat consistency as a creative asset
Character sheets, reference architecture, prompt design, and model selection are becoming foundational techniques in AI filmmaking. But consistency extends beyond characters. It can also apply to visual language, design systems, style, and proprietary workflows.
Know how to collaborate
AI creators can increasingly complete projects alone. That does not mean solo production is always stronger. The ability to understand the strengths of other creators and established-industry specialists, divide responsibility quickly, and build toward one result becomes especially visible under time pressure.
Speed alone will not differentiate you
Generation will continue to become faster, with real-time possibilities increasingly entering the conversation. The differentiator is not what you managed to generate in 50 minutes, but what you chose in those 50 minutes. As speed becomes commoditized, direction and judgment become more valuable.
Story, direction, and taste matter more
CRE[AI]TE’s judging criteria point toward a future in which AI filmmaking is evaluated through story, originality, aesthetic judgment, sound, and execution. Model literacy still matters, but “I used the newest model” will not remain a meaningful creative proposition for long.
Distribution, community, and marketing are part of the work
Great work cannot create opportunities if it never reaches the right people. Production ability increasingly needs to be paired with creator networks, platform participation, consistent publishing, community connection, and an understanding of how the work reaches an audience.

From the Japanese creator ecosystem specifically, the larger challenge is connection rather than craft alone: English-language communication, global communities, international events, team production, media visibility, and brand collaboration.

For AI Creators as a platform, the lesson is similar. Simply listing creators is not enough. We need infrastructure that connects their expertise, creative work, and original IP with opportunities in Japan and internationally.

CapCut World Showed the Future of the Creator Economy, Not Just AI Tools

What became visible at CapCut World 2026 Singapore was not simply a list of new AI features.

Templates circulate finished structures. Skills can make creative process reusable. AI Lab, Infinite Canvas, generation, Editor, and Edit Pilot provide the execution environment. A global creator community connects the people using those systems. And three distinct recognition mechanisms evaluate creators, finished work, and live collaboration in different ways.

Bringing those elements together in one event made CapCut’s direction easier to understand than a feature list alone could.

In AI creation, the ability to generate will gradually become less valuable as a differentiator. Models improve. Generation gets faster. Baseline quality rises.

What matters next is who is making the work, what they are making, why they are making it, and how they choose to make it.

And then comes another question: who they share the method with, and which community carries it into the next project. CapCut World Singapore was significant not simply as a preview of AI tooling, but as a working sketch of the creator economy those tools are being built into.

Los Angeles is scheduled to follow Singapore, then London. Different cities will bring different creators, different work, and different creative cultures. What remains consistent — and where the format begins to change — will be worth watching.
(Sources: CapCut World 2026 / CapCut Newsroom)

FAQ

What is CapCut World 2026?: CapCut World 2026 is CapCut’s first global community event series, with editions in Singapore, Los Angeles, and London. It combines product experiences, creator workshops, screenings, awards, and public-facing creative programs. (Source: CapCut World 2026)
When and where was CapCut World Singapore held?: The public program ran at Marina Bay Sands from September 4–6, 2026. The invitation-only main event at Capella Singapore took place on September 5, 2026. (Source: CapCut World Singapore | CapCut Japan)
What is CapCut Skills?: CapCut Skills is a feature designed to save, reuse, and share styles, workflows, and techniques behind creative work. While Templates primarily make finished structures reusable, the Skills feature extends reuse to the process behind the work. It rolled out in Design Studio on August 10, 2026, ahead of CapCut World Singapore.
What is CapCut CRE[AI]TE?: CRE[AI]TE is CapCut’s first AI Festival, an international competition launched in 2026. It includes Film, Series, Creative, and Commercial categories and recognizes 30 works in total: 10 Cash Awards and 20 Spotlight Selections. (Source: CapCut CRE[AI]TE 2026)
What is the CapCut AI Creators Hackathon?: The CapCut AI Creators Hackathon was a team-based, time-constrained production challenge held within the invitation-only CapCut World Singapore program. The author’s team created the 90-second short film ROOM 0 in approximately 50 minutes and received Best Main Character Design.
What is the difference between Creator Awards, CRE[AI]TE, and the Hackathon?: They recognize different forms of creative value. Creator Awards focus on creator activity and contribution, although the detailed selection methodology was not disclosed in the public materials reviewed for this article. CRE[AI]TE evaluates finished work through published judging criteria and a jury. The on-site hackathon evaluated live team production, with participant voting after all works were screened.
How many people attended CapCut World Singapore?: According to CapCut, more than 300 creators, industry professionals, and CapCut team members attended in Singapore. The number does not refer to creators alone. (Source: CapCut Newsroom)
Which AI tools did CapCut highlight at CapCut World Singapore?: CapCut highlighted a workflow involving AI Lab, Infinite Canvas, Edit Pilot, and related creative environments, alongside AI models including Dreamina Seedance for video, Dola Seedream for images, and Seed Audio for sound. CapCut says it now offers more than 85 AI tools. (Source: CapCut Newsroom)
What is ROOM 0?: ROOM 0 is a 90-second short film created at the CapCut AI Creators Hackathon in Singapore. Four creators brought original characters into a single story built around a deadly countdown and a series of game-like challenges. The film was completed in approximately 50 minutes and received Best Main Character Design.
Where is CapCut World going next?: After Singapore, CapCut World is scheduled to continue in Los Angeles in October 2026 and London in November 2026. (Source: CapCut World 2026)

en.ai-creators.tech

Generative AI Courses & AI Creator Network

https://en.ai-creators.tech/personal/

Learn practical generative AI skills, build your portfolio, or apply to join our global AI creator network and connect with new creative opportunities.

en.ai-creators.tech

Generative AI Creative Services for Business

https://en.ai-creators.tech/enterprise/

AI Creators supports companies with generative AI creative production, AI creator matching, corporate training, strategic advisory, PR, and partnerships.

The post CapCut World 2026 Singapore: Skills, Hackathon & Creator Awards first appeared on AI Creators.

AI Is Becoming a Visual Language | Minouche Hijkoop Interview

aratama 璞 | Editor-in-Chief, AI Creators — Mon, 10 Aug 2026 00:14:03 +0000

Generative AI has made it easier than ever to produce a striking image or a beautiful shot. But being able to generate something beautiful is not the same as building a world that stays with an audience.

What AI creative work increasingly requires is not prompt technique alone. It calls for a visual language, emotional design, character consistency, narrative clarity, and an identity that runs through the whole piece.

For this feature, AI Creators spoke with Minouche Hijkoop, a Netherlands-based AImaginist and VFX Prompt Designer. Minouche is the founder of the AI cinematic-identity studio Shine by Nous and co-founder of PHI, a project she created with the Japanese AI creator Kasamacura. As one of the first twenty Adobe Firefly Ambassadors worldwide, and working across more than eighteen major AI platforms, she explores what AI-driven film, character design, and cinematic identity can become.

The interview is conducted by aratama 璞, an AI Creative Architect / Artist, who explores the core of Minouche’s thinking: what it means to treat AI as a visual language, and where AI-native production is heading.

*This article is based on a written interview with Minouche Hijkoop. Her answers appear in her own words, lightly edited for length and flow.

Profile

Minouche Hijkoop | AImaginist / VFX Prompt Designer

Minouche Hijkoop is a Netherlands-based AI Artist / AImaginist / VFX Prompt Designer. She is the founder of the AI cinematic-identity studio Shine by Nous and co-founder of PHI, a project she created with the Japanese AI creator Kasamacura.

Her work centers on high-impact AI image and film production, spanning visual storytelling, style development, cinematic identity, character systems, and VFX prompt design.

As one of the first twenty Adobe Firefly Ambassadors worldwide, she works with a range of major AI platforms—testing new features, providing feedback on user experience, and exploring how AI fits into professional production. Her work was selected for the AI Film Awards Cannes 2025 and named a finalist at the Bali International AI Film Festival 2025.

en.ai-creators.tech

Shine by Nous

https://en.ai-creators.tech/creators/shinebynous/

AI Artist | Creative Partner with Leading AI Tools | VFX Prompt Designer. I’m Minouche Hijkoop (Netherlands), founder of Shine by Nous and co-founder of PHI with Kasamacura (Japan).

Chapter 1: Where an AI Creator Really Comes From

Minouche’s path did not begin in art or film. Before generative AI, she worked as a technical manager in building services engineering—handling structure, planning, cost calculations, project proposals, and the technical problems clients brought to her.

We tend to picture AI creators as coming from video, design, or art. Minouche’s background suggests something else: that creativity in the AI era can also grow out of engineering, management, problem-solving, craft, and comedy—fields that look, at first, unrelated.

aratama 璞

To start, tell us about your background before you moved into this work. How does your experience before generative AI feed into the visual work you make now?

Minouche Hijkoop

If someone had told me a few years ago that I would work in AI, I probably would not have believed them.
At the time, I worked as a technical manager in building services engineering. My job involved structure, planning, cost calculations, project proposals, and solving technical problems for clients.
Accuracy and attention to detail were essential, because one small mistake could affect the entire outcome.

Collaboration was equally important. In 2021, my team was named the Netherlands’ Superteam during De Slimste Handen van Nederland. That recognition was based on trust, communication, and solving complex challenges together.

Outside work, I designed personalized clothing with heat transfer vinyl and rhinestones, and I wrote and performed stand-up comedy. AI unexpectedly brought those different experiences together—technical precision, visual design, timing, and storytelling now all shape my creative process.

When health problems changed my career, generative AI opened a completely new path. It became the medium where technology, creativity, and storytelling finally came together.

For Minouche, AI was never just another production tool. It became a language that reconnects experiences that used to sit apart. In the AI era, a cross-disciplinary career is not a detour to be discarded—it becomes raw material for judgment. Engineering precision, comedic timing, the maker’s eye from clothing design: the real question is how to re-edit all of that inside a single medium. That is where the entry point to AI creative work turns out to be more than “adapting to a tool.”

Chapter 2: What “AImaginist” Actually Means

Minouche describes her own role as AImaginist—a word for a practice that “AI Artist” or “Prompt Engineer” alone could not capture.

aratama 璞

You describe yourself as an AImaginist. What do you mean by that term?

Minouche Hijkoop

For me, an AImaginist is someone who uses AI to turn imagination into visual reality. I created the term because no existing title fully describes what I do. My work combines art, technology, storytelling, direction, visual research, worldbuilding, and production.

I do not begin with a tool. I begin with an idea, an emotion, or a world that does not yet exist. I then explore how to make that vision visible: how a character looks and moves, which lighting creates the right mood, which camera angle strengthens the emotion, and how separate images can become one coherent story.

Prompts are only one part of that process. Selection, iteration, direction, consistency, and post-production are just as important.

I therefore see an AImaginist less as one fixed profession and more as a creative connector. It describes someone who brings human imagination and generative technology together. The tools may change, but the purpose remains the same: making visible what previously existed only in the mind.

The heart of her approach sits in one line: “I do not begin with a tool.” In the generative-AI space, the conversation usually leads with model names, features, prompts, and updates. But the thing that makes a work hold together starts on the other side—with emotion, with a question, with a discomfort that has not yet found words. AImaginist is, in part, a way of pulling the starting point of creation back from the tool and toward the imagination.

Chapter 3: AI as a Language, Identity as Signal

One line from Minouche’s profile stays with you:

AI as a language. Identity as signal.

Treat AI not as a mere generator but as a new visual language; and treat the creative signature that runs through a body of work as its identity. These two ideas are the foundation of how she works.

aratama 璞

“AI as a language” and “Identity as signal” read like a very deliberate statement. What does it mean, for you, to treat AI as a visual language?

Minouche Hijkoop

For me, AI is not a replacement for creativity. It is a new visual language for making ideas visible. Like photography, film, or animation, generative AI has its own grammar.

You learn more than prompts. You learn rhythm, composition, camera logic, lighting, movement, and emotion. The better you understand that language, the more clearly you can communicate a creative vision.

“Identity as signal” means the creative signature that can be felt throughout the work. It appears in the way a scene moves, the choice of colors, the camera angles, the atmosphere, and the rhythm of the edit. Sometimes you can recognize a creator’s work without seeing their name. That recognition is identity.

This is why I rarely create from a single prompt. I build systems in which characters, environments, cinematography, and style support one another. When these elements speak the same language, the result becomes more than a collection of beautiful images. It becomes a world with its own character and a recognizable visual identity.

Generative AI has sharply raised the speed at which anyone can reach a high-quality image. That is exactly why the value of a work shifts from one-off beauty to authorship. “Identity as signal” names that authorship precisely. What you choose again and again, what you avoid, what rhythm and air you leave across the whole piece—as AI levels more of the craft, this consistent sum of decisions becomes the one signal that sets a creator apart.

Chapter 4: Cinematic Identity Is the Personality of a Story

A central theme in Minouche’s work at Shine by Nous is AI-driven cinematic identity. It is not about making something look filmic; it is about designing the personality of a story—its emotional flow, its visual language, its consistency of style.

aratama 璞

What does cinematic identity mean to you? How do you build a visual world that feels genuinely alive rather than generic?

Minouche Hijkoop

For me, cinematic identity is the personality of a story. It begins with emotion, not with a prompt or a tool. Before creating anything, I first ask what the audience should feel. Only then do I begin shaping the visual world around that emotional goal.

Cinematic identity is built through many connected choices: color, lighting, camera work, composition, rhythm, textures, characters, and movement. Individually, these may seem like small details, but together they determine whether a world feels believable or generic.

I spend a great deal of time researching references, testing styles, and building a visual language in which every element strengthens the others. My aim is for viewers to recognize the atmosphere or style before they even see a title or logo.

Generative AI changed what felt possible for me. It brought technology, storytelling, and imagination together in a way that felt natural. It did not feel like discovering only a new tool. It felt like discovering a creative language that finally matched the way I think.

The detail worth noticing is the order of work. Most AI production starts from “make a good image.” Minouche starts by fixing the emotional goal—what the audience should feel—and works backward from there to color, light, camera, and texture. Cinematic identity is not the beauty of any single look; it is what emerges when every visual decision is aligned in one direction from an emotional foundation. Reverse that order, and the world falls back into surface-level prettiness.

Chapter 5: VFX Prompt Design Is Creative Direction

“Prompt” is often heard as the single line you type into an AI. For Minouche, VFX prompt design means something wider: direction across the whole of production—camera, lenses, light, texture, scale, physics, emotion, and post.

aratama 璞

As a VFX Prompt Designer, how do you think about prompts? What separates this from simply writing a good sentence?

Minouche Hijkoop

VFX Prompt Design is very different from simply writing one good prompt. The term “prompt engineer” can suggest that the work is about finding a perfect sentence, but in reality the process often begins after the first generation.

For me, VFX Prompt Design is a form of creative direction. I think about camera movement, lenses, composition, lighting, materials, scale, physics, continuity, and emotional tone. I also need to understand how different AI models behave, because the same idea may require a completely different approach in another tool.

A strong prompt system depends on iteration. I analyse results, adjust details, and often test many versions before a scene feels right.

I also consider how each shot will function in the final production. Post-production remains essential for compositing, motion graphics, refinement, fixing imperfections, and creating consistent movement. AI provides important building blocks, but the professional result comes from combining AI, direction, craftsmanship, and traditional production tools within one coherent workflow.

What she describes is a view in which the prompt is an entrance, not the work. The stages that begin after the first generation—deciding what to keep and what to cut, iterating, compositing, fixing, connecting one shot to the next—are what set the final quality. In VFX especially, generating a flashy effect is not enough to make a shot feel cinematic. Eye-line, the visual weight inside the frame, continuity between shots, the link to sound and edit: only when those are designed does a sequence become a film. AI is a powerful supply of material, but the direction that binds it stays on the human side.

Chapter 6: Turning Fragmented Ideas into Production-Ready Direction

AI makes the early visualization of ideas dramatically faster. But being able to output quickly is a different thing from converting fragmented moods and concepts into a coherent visual system.

aratama 璞

Many creative teams have a mood, a concept, an emotional direction—yet not a full visual world. How do you translate those fragments into production-ready visual direction?

Minouche Hijkoop

I work strongly from emotion, so before generating any images, I first try to understand what a scene or story should make the audience feel. That emotional intention becomes the foundation of the visual direction.

Many clients or creative teams begin with fragments: a mood, an idea, a reference, or a feeling they want to communicate. My role is to connect those fragments and turn them into a coherent visual system. This often starts with developing a story, script, moodboard, or storyboard in which every choice supports the intended emotional impact.

AI is extremely valuable in the early stages of a project because ideas can be explored, tested, adjusted, and presented much faster. However, the technology does not replace the creative translation itself. You still need to understand storytelling, psychology, and emotional nuance, and you must be able to direct the tools clearly.

Used well, AI helps transform an abstract concept into a visual world that can support pitching, creative development, planning, and eventually full production.

The key term is Minouche’s own: creative translation. AI shows ideas quickly, but speed does not guarantee depth. Line the fragments up as they are and you get a collage of mood boards. The work is to bridge scattered impressions across different layers—emotion, story, image, production process—and bind them in a single direction. That translation is becoming a core competence for the director or producer of the AI era.

Chapter 7: Making Characters and Worlds That Feel Alive

An AI-generated character needs more than an appealing face. How it moves through a story, what it feels, how it changes: without that continuity, a character stays a beautiful still.

aratama 璞

You talk about building story-based worlds and character systems rather than generic AI output. What does it take for an AI-generated character or world to feel alive?

Minouche Hijkoop

For me, AI creation is not mainly about producing individual images. What interests me most is building complete stories, cinematic productions, and visual worlds that feel alive.

A character needs more than an attractive face or design. It needs personality, emotion, body language, and recognizable behavior. Before generating images, I think about who the character is, how they move, what they feel, and how they develop throughout the story. These choices create the foundation for visual consistency.

The same applies to the world around the character. Lighting, color, cinematography, architecture, sound, atmosphere, and pacing all need to support one another. Only then does the world feel believable.

Consistency remains one of the greatest challenges in AI production. Maintaining the same character, style, and atmosphere across many scenes requires a structured workflow, constant iteration, and often post-production.

For me, AI is simply a tool. The real goal is immersion. When the audience forgets how the work was created and becomes absorbed in the story, the world has truly come to life.

In AI character work, the attention usually falls on one kind of consistency—”can you hold the same face?” Minouche is asking about the step beyond it: is the character behaving as the same being? Expression, timing, reaction, growth, the link to sound—as those accumulate, a character turns from an interchangeable visual asset into a specific presence an audience remembers. AI-native character work is expanding beyond exterior design into the design of behavior and emotional continuity.

Chapter 8: There Is No Perfect AI Tool

Minouche works across more than eighteen AI platforms—Adobe Firefly, Runway, Kling, HAILUO.AI/Minimax, Wonder Studios, and others. What that experience reveals is a plain reality: there is no perfect tool.

aratama 璞

As a creative partner across several major AI platforms, what makes an AI tool genuinely valuable for professional production?

Minouche Hijkoop

Working with more than eighteen AI platforms has taught me that there is no perfect tool. Every platform has different strengths and limitations, so I focus less on popularity and more on which tool fits a specific stage of production.

For professional creators, model quality alone is not enough. A valuable tool must be reliable, offer meaningful creative control, and fit smoothly into an existing workflow. Important features include consistent results, strong character references, accurate camera control, flexibility, and an intuitive user experience.

Creative partners can help because we use these tools in real production situations. We see where workflows fail, which features save time, and which limitations create problems for creators. That practical feedback can help companies build better products, not only better models.

Responsible development is equally important. Transparency, respect for creators, and serious attention to community feedback are essential. The best innovation happens when developers and creators work together, with both technical progress and real creative needs in mind.

Tool evaluation often leans on image quality and generation speed. But what the field actually asks is not raw model performance—it is character reference, camera control, stability, how well a tool slots into team production, and how it connects to post. Only when these line up does AI move from “experiment” to “production environment.” And what accelerates that shift is feedback from creators who feel the friction in real production.

Chapter 9: Adobe Firefly and Responsible Creative AI

As one of the first twenty Adobe Firefly Ambassadors worldwide, Minouche has followed Firefly’s development from an early stage—a vantage point on how generative AI moves from experimental technology into professional creative workflows.

aratama 璞

You’ve watched generative AI enter professional production up close as an Adobe Firefly Ambassador. What should responsible, creator-friendly AI look like?

Minouche Hijkoop

As one of the first twenty Adobe Firefly Ambassadors worldwide, I have been able to follow Firefly’s development from an early stage. This gave me valuable insight into how generative AI is becoming part of professional creative workflows rather than remaining only an experimental technology.

For professional use, trust is essential. Artists, brands, and production teams need reliable tools, transparency, and workflows that connect smoothly with existing creative software. AI should support creativity, not replace it or force creators into disconnected systems.

One of my most memorable experiences was attending Adobe MAX in Los Angeles in October 2025. Meeting other ambassadors in person reminded me that behind every AI tool are real people: artists, developers, designers, filmmakers, and creators working toward meaningful progress.

Responsible AI also requires listening to the people who use these tools every day. Their feedback helps improve usability, creative control, accessibility, and production quality.

Creator-friendly AI should remove technical barriers and give people more time for storytelling, emotion, and ideas. That is where I believe AI can make the greatest difference.

Discussion of AI tends to emphasize efficiency and automation. Minouche points the other way. AI is most valuable not when it takes creativity away from people, but when it gives them more time to spend on emotion, story, and ideas. The more a tool absorbs the technical barriers, the more a creator can concentrate on judgment and expression. Responsible AI, in that sense, is a design philosophy that protects this reallocation of time.

Chapter 10: Can AI Film Move Beyond Technical Spectacle?

Minouche’s work was selected for the AI Film Awards Cannes 2025 and named a finalist at the Bali International AI Film Festival 2025. The international AI-film scene is expanding fast—and still in transition.

aratama 璞

How do you see the international AI-film and AI-creative scene right now? What will it take for the field to mature beyond technical spectacle?

Minouche Hijkoop

It is a great honor that my work was selected for the AI Film Awards Cannes and that I was a finalist at the Bali International AI Film Festival. These festivals show how quickly AI is developing as a creative medium and how many creators worldwide are exploring new forms of storytelling.

At the same time, the field is still in transition. Much attention remains focused on technical spectacle: impressive images, complex prompts, and dramatic effects. That is understandable, but audiences do not remember the technology itself. They remember the story and the feeling it leaves behind.

The work that stands out to me is not always the most technically complex. It is the work in which the technology becomes almost invisible and fully serves the characters, emotion, and narrative.

I believe the next stage will depend on stronger collaboration between filmmakers, writers, editors, VFX artists, sound designers, and AI specialists. As the technology becomes more stable, the focus should shift from how something was made to why it was made. That shift will help the field mature.

In its early stage, AI work draws attention to “how it was made”—which tool, which prompt, how hard the technique was. As a culture matures, the question moves to “why it was made” and “what it leaves behind.” Whether a work survives once the technology becomes invisible—as story, emotion, and character—is the test. More work that can withstand that question is what moves AI creativity from a passing trend toward a lasting creative culture.

Chapter 11: PHI, Japan, and Cross-Cultural Collaboration

Minouche co-founded the project PHI with the Japanese AI creator Kasamacura. It began not from a business plan but from a shared creative vision.

aratama 璞

Tell us about PHI, the project you co-founded with Kasamacura in Japan. What vision did it grow from, and how has the collaboration with Japan shaped your work?

Minouche Hijkoop

PHI was born from a shared creative vision rather than a business plan. Kasamacura and I both wanted to create original work without being limited by trends or conventions. That shared freedom to experiment became the foundation of PHI.

Our styles are very different, and that is exactly why the collaboration works. We do not try to imitate or control each other. Instead, we encourage one another to explore ideas freely.

We have also created a shared universe in which Kasamacura’s characters, including Liv, Alice, Izarin, and XIMENA, can interact with my character Mogg. Mogg is a stubborn little green elf who believes he is always right, while almost everything he does goes hilariously wrong. We want to develop these characters into larger stories and a comedy series.

People sometimes expect Japan to have changed my visual style, but the biggest influence has not been one specific aesthetic. It has been the creative freedom of working across cultures. PHI is not about making Dutch or Japanese work. It is about creating something neither of us would have made alone.

International collaboration in AI creative work is not about blending nationalities or cultures. What makes PHI interesting is that Minouche and Kasamacura do not force their very different styles into one. They keep the difference intact and move toward expression neither could reach alone. Now that AI lets characters, worlds, images, and stories connect more flexibly, cross-cultural collaboration is shifting its weight from “mixing” toward “resonating.”

Chapter 12: The Future of AI-Native Production

Three to five years is hard to predict in AI. Even so, Minouche sees a direction: away from AI standing out as a special category, and toward AI integrated deeply into ordinary creative production.

aratama 璞

Looking three to five years ahead, how do you think AI-native visual production will develop? And what work or expression do you want to explore next?

Minouche Hijkoop

Three to five years is a long time in AI, so I am cautious about making specific predictions. I expect AI to become less visible as a separate category and more integrated into regular creative production. The main question will no longer be whether AI was used, but whether the result serves the story, the audience, and the purpose.

Creative roles will continue to evolve. Individual creators may develop ideas further on their own, while professional productions will still depend on collaboration between writers, directors, editors, VFX artists, sound designers, and AI specialists. Human judgement, taste, emotion, and experience will remain essential.

Through Shine by Nous, I want to continue creating cinematic, story-driven productions with brands and creative teams that value quality. With PHI, I want to expand our shared universe and further develop Mogg as the foundation for a larger comedic world.

I am also investing in workshops, courses, and learning from experienced professionals. Combining experimentation with deeper knowledge will help me create stronger and more purposeful work. Above all, I want to keep making work that I genuinely want to see.

“Not whether AI was used, but whether the result serves the story, the audience, and the purpose”—this is the measure for the maturity of AI-native production. The moment when AI draws attention as a special technology will pass. What remains is not a tool’s name but what a work makes people feel and what world it leaves in their memory. Minouche’s closing line—that she wants to keep making work she genuinely wants to see—is that same measure, put in its most personal form.

Authorship and the Production Environment of AI Creativity

What emerges from Minouche Hijkoop’s words is that AI creativity is not simply tool use. It is moving toward a broader creative environment—one that includes visual language, production design, character continuity, international collaboration, and responsible tool development.

“AI as a language” pulls AI away from being a means of efficiency. Like photography, film, or animation, it is an expressive language to be learned, handled, and developed. What you say in that language—what emotions and worlds you raise with it—is the question now put to AI creators.

And “Identity as signal” names authorship in the generative-AI era. The more anyone can generate a high-quality image, the more a work’s value lives not in single beauty but in the judgment, structure, rhythm, emotion, and visual signature that run through the whole.

At AI Creators, our aim is not only to introduce people who can use AI, but to surface each creator’s own visual language, production philosophy, worldview, and possibilities for collaboration—and to connect them to the next opportunity to make something. The future of AI-native expression is not decided by technical progress alone. It is shaped by human imagination, judgment, emotion, and experience, and by production environments where different talents meet.

Minouche’s practice offers one clear vantage point on that future. AI is not a replacement for creativity. It is a new language for turning what is not yet visible into something you can see.

From Introductions to Real Projects

AI Creators works to discover and introduce creators opening new forms of AI expression, to support their production, and to build collaborations with companies and organizations.

If you’re interested in AI-driven film production, character development, visual identity, advertising, content production, or international collaboration, explore the entry point that fits your goal.

en.ai-creators.tech

Generative AI Courses & AI Creator Network

https://en.ai-creators.tech/personal/

Learn practical generative AI skills, build your portfolio, or apply to join our global AI creator network and connect with new creative opportunities.

en.ai-creators.tech

Generative AI Creative Services for Business

https://en.ai-creators.tech/enterprise/

AI Creators supports companies with generative AI creative production, AI creator matching, corporate training, strategic advisory, PR, and partnerships.

The post AI Is Becoming a Visual Language | Minouche Hijkoop Interview first appeared on AI Creators.

International AI Creator Dialogue: Victor Moreno × aratama 璞 — AI-Native Expression and the Future of Production

aratama 璞 | Editor-in-Chief, AI Creators — Fri, 12 Jun 2026 01:01:37 +0000

Generative AI is rapidly lowering the barriers to entry in filmmaking.
Visual work that once required substantial budgets, camera equipment, and specialized production teams can now be explored by individual creators within a much shorter timeframe.

Yet being able to generate eye-catching footage is not the same as creating work that stays with an audience.
The more capable the models become, the more the questions circle back to the fundamentals.
What are we making? Why are we making it? How do we edit it, layer sound into it, and shape it into a coherent experience?

And precisely because AI creators can now move across so many disciplines on their own, the question of who they collaborate with — and how — is becoming increasingly important.
For this feature, AI Creators invited Stockholm-based Victor Moreno and Tokyo-based aratama 璞 to take part in a written dialogue.

*This article is based on a written interview with Victor Moreno and responses from aratama 璞. Moreno’s comments have been lightly edited for length and clarity without altering their intended meaning.

Profiles

Victor Moreno｜Stockholm, Sweden

photo credit: Catherine Ward

Victor Moreno is an AI creator and film director based in Stockholm, Sweden.
With a background spanning graphic design, art direction, editorial photography, and cultural journalism, he works across AI-driven filmmaking, brand expression, and narrative projects.
His clients include Telia Sweden, Netflix, Lunar Bank, and Suicide Zero.

His short narrative film Hazama (狭間) has been selected for AI Film Festival Japan, Cinema Shift Toronto, and AI Media Festival (AIMF) Los Angeles.
In 2024, he joined Runway’s Top 50 Talent Network and became a collaborator with generative AI company Luma Labs.

en.ai-creators.tech

Victor Moreno

https://en.ai-creators.tech/creators/victormoreno/

Victor Moreno is an AI Director creating cinematic brand films, corporate films, and narrative works with generative AI and strong visual authorship.

aratama 璞｜Tokyo, Japan

aratama 璞 is a Founder / AI Creative Architect / Artist working across generative AI, moving images, music, live performance, and spatial experience.
His practice spans character design, virtual humans, music videos, advertising, event visuals, and audiovisual works, integrating concept development, direction, experimental R&D, and production from end to end.

In recent years, rather than presenting moving-image works solely as finished pieces, aratama has been exploring ways to connect music, visuals, space, physical presence, and audience experience through an audiovisual practice using AI.
Through AI Creators, he is also involved in discovering and developing AI creators, supporting production, building connections with companies, and establishing new criteria for evaluating creative work.
Rather than treating AI simply as a tool for efficiency, he focuses on designing new forms of expression that move across established creative disciplines.

en.ai-creators.tech

aratama 璞

https://en.ai-creators.tech/creators/aratama/

Founder, Producer, Artist. I work on projects, exhibitions, event appearances, and workshops utilizing generative AI and web3 both domestically and internationally.

Chapter 1　AI Has Opened the Door to Filmmaking — But Quality Does Not Improve Automatically

Victor Moreno sees the democratization of filmmaking through AI first and foremost as a practical reality.
Much like the transition from film to digital, AI reduces many of the major costs once associated with film processing, camera equipment, lighting, transportation, and production crews.
As the barriers to entry fall, more people gain the opportunity to explore visual storytelling.

But that openness can also be a double-edged sword.

Victor Moreno

Generative AI is lowering the barriers to filmmaking.
The result will be plenty of noise and a lot of mediocre work.
But some beautiful work will emerge from it as well.

What matters is not simply being technically impressive.
The work needs to speak to an audience.
It needs to resonate with the person receiving it.

The Sex Pistols were not musical virtuosos.
But they had a purpose that connected with the cultural and societal moment.

Working with AI is not the point in itself.
What matters is how you incorporate those new possibilities into the creative process and the production pipeline.

aratama 璞

AI has made it possible for individuals to experiment with forms of visual expression that once required specialist production companies.
But being able to generate something and being able to create a fully realized work are not the same thing.

As the number of works increases, the weight carried by each individual piece can easily diminish.
It is no longer enough simply to make something.
We need to ask why the work exists, how it should be presented, and how it can remain in the audience’s memory.
Another major shift is that we can now devote more time to refining the intention and presentation of a work, rather than focusing solely on the labor of production.

In a world where the barriers to entry have fallen, the rarest quality is not the ability to generate content.
It is the ability to ask the right questions, form a clear intention, think critically, and make decisions.

Chapter 2　“The Model Is Just the Instrument” — What Makes Work Last Is Intention and Direction

Generative AI tools are evolving rapidly.
But as models become more capable, the ability to use a particular tool becomes less meaningful as a point of differentiation.

The question is what a creator can draw out of the model and how those elements are shaped into an experience.
This is where an individual’s accumulated skill and experience come into play.

Victor Moreno

Creative direction, taste, and intention are everything.
The model is just the instrument.
Working with AI is about taming the beast.
Owning it, bending it toward a vision.

Lately, I’ve been drawn to the word “sampler” as a way of describing how generative AI works, and I think it is accurate.
Some of my early work was built using Flux Redux, which is exactly that: taking elements, sampling them, shaping and modulating them until the result reflects what you are actually after.

That craft is no different from what musicians, directors, and artists have always done.
A lot of the most iconic work in pop culture was built on sampling.
The raw material was borrowed, but the vision behind it was entirely original.

aratama 璞

When people discuss AI-generated work, the model or the generation technique often comes first.
But using the same model does not produce the same work.

What matters is not simply receiving an output, but deciding what to keep, what to discard, how to sequence it, and how to transform it into an experience.
The value of an AI creator lies not in the number of generations they produce, but in the quality of their decisions.

Victor’s metaphor of the instrument captures this perfectly.

Chapter 3　The Fundamental Challenges That Still Remain in AI Filmmaking

AI filmmaking has advanced dramatically within a short period of time.
At the same time, essential challenges remain: consistency between shots, control over longer sequences, character continuity, natural emotional expression, and narrative structure.

Victor Moreno

Since day one of working with AI, consistency and continuity across shots, and control over longer sequences, have been the main challenges.
We are in the middle of 2026, and I have only just started working with character emotional depth, which is getting better and better as well.
Progress is extremely fast.

What is a limitation today becomes a solution tomorrow.
That is one of the great things about working with AI.

When you know what you are after, you explore the possibilities and execute.
Star Wars used matte paintings — hand-painted on plexiglass.
Working with AI is no different.

If you want to maximize the outcome, I do not think you can rely on text-to-video alone.
You want to incorporate motion capture, LoRA training, video-to-video — whatever gets you closer to the vision.

And still, I have seen beautiful accidents come from a single generation.
But you need to lift the raw material: audio sync, SFX, music, editing pace, grading — something that elevates it into a work that actually lands with an audience.

aratama 璞

In AI filmmaking, individual shots can be striking while the work as a whole feels weaker once those shots are placed in sequence.
Moving images are not simply a collection of stills.
They are an experience shaped by music, rhythm, pauses, speed, structure, and changes in space.

Improvements in model performance matter, but ultimately the question is how the entire work is designed.

How do we place the beauty of an accidental frame, or movement that cannot be fully controlled, within the structure of an experience?
That is where the essence of editing lies.

Chapter 4　AI Filmmaking Quality Depends on Workflow, Not Just the Model

Building longer, more polished sequences requires the design of an entire production process, not simply the accumulation of individual generations.
Reference images, storyboards, shot breakdowns, compositing, editing, sound, and color grading.
The accumulation of decisions made around the model shapes the final viewing experience.

Victor Moreno

For me, it comes down to reference design and editing pace.
Those are the anchors.

There is an unspoken nature to what the brain identifies in a blink.
Your job as a storyteller is not to alienate your audience with invisible elements that make them drift — to start thinking about what time it is, or what they need to do tomorrow.
Once that happens, they have already left the work.

Consistency, rhythm, dialogue, sound, grading.
Those are your allies in delivering a well-packaged story.

The real craft is knowing what you do not need to put in the frame.
Letting the audience build that part on their own.
That is where the magic happens.

Do I think the quality of AI filmmaking will increasingly depend on the production workflow around the model rather than the model alone?
Absolutely.

That said, I can also foresee genres or subgenres emerging where the raw model output becomes the aesthetic itself.
Almost like lo-fi became a genre rather than a limitation.

aratama 璞

AI rapidly increases the number of elements we are able to add.
But making a work stronger requires not only addition, but bold subtraction.

What should remain unseen?
At what point should the image stop?
How much explanation should remain?

The more possibilities AI creates, the greater the weight placed on the judgment of directors and editors.
Victor’s observation about knowing what not to put in the frame becomes even more important in an era of abundant tools.

Chapter 5　Can One AI Creator Carry an Entire Production?

AI creators can now work across a much wider range of processes than before.
Planning, concept development, visual direction, prompt design, compositing, editing, sound, narrative structure, workflow design, and distribution.
The scope of what a single creator can manage has undeniably expanded.

But that does not mean every creator should be expected to complete everything alone.

aratama 璞

AI creators can now take on many different disciplines.
But to raise the quality of the work further, I believe we need specialist teams that understand AI.
Victor, how do you see the role of the AI creator?

Victor Moreno

I see an AI creator as a director, like an orchestrator.
AI can help you push things 50, 60, even 70 percent of the way.
But that final stretch is the longest and most demanding.

Having the support of a specialized editor, sound engineer, or colorist makes a measurable difference.
Two years of working full-time with AI have shown me that.
Even in the early stages, I have been incredibly grateful to have a team around me: technical support for workflow troubleshooting, AI artists handling image retouching, copywriters, editors, and all the post-production craft that elevates the work.

What I find striking in 2026 is that large corporations still have not moved to build these teams in-house.
There is a widespread misreading of what AI filmmaking actually is — the assumption that anyone can do it.
When in-house attempts fall short, people blame the technology rather than the lack of craft and structure around it.

The US is probably ahead.
I have seen some rare opportunities in the UK as well, but they still lean heavily on engineering and technical skills: Python, workflow builders.

That makes sense when you are trying to create something scalable and sustainable.
But the expertise required to execute work as moving-image expression is just as important.

aratama 璞

AI creators sometimes carry a wider range of responsibilities than creators in more established production fields.
They can handle planning, direction, generation, editing, sound, implementation, and distribution alone.
But that breadth can also become a limitation.
Talented AI creators are often occupied with existing commissions and have too little time to develop their most ambitious work.

The important question is not simply who is talented.
What are that creator’s strengths?
What kind of support would allow the work to reach the next level?
What specialists, technologies, funding, time, and production environment are required?
The answer will differ from creator to creator.

In some cases, collaboration between AI creators may be the right answer.
In others, the work may require connections with established specialists: producers, screenwriters, editors, sound designers, colorists, engineers, animation specialists, 3DCG designers, marketers, and industry advisors.

Truly powerful AI-native work emerges when an individual vision is supported by the right collaborators, production resources, time, and funding.
For AI Creators, the challenge is not simply to showcase talented creators, but to consider how their abilities can be developed, how the right teams can be assembled, and how the work can be elevated to the next level.

Lunar Bank – “Old Bankers” TV Commercial (2025)

Client: Lunar Bank
Production company: UNCUT
Executive Producer: Fredrik Skoglund
AD: Tor Söderholm and Victor Moreno
Copywriter: Erik Bergqvist
AI team: Ib Thorub, Lars Bjurman, Victor Moreno, Simon Appel
Editor: Karim Fakih
Colorist: Filip Bergh

Chapter 6　Selection at AI Film Festival Japan and the Potential Victor Sees in Japan

Victor Moreno’s narrative film Hazama (狭間) was selected for AI Film Festival Japan.
Although he was unable to attend the screening at Tokyo Innovation Base, he describes learning that Nippon Television had attended the event and broadcast footage from the film on NTV News as an especially meaningful moment.

Victor Moreno

I genuinely regret not being able to attend the screening at Tokyo Innovation Base.
But learning that Nippon Television visited the venue and broadcast footage from my short film Hazama (狭間) on NTV News was incredibly rewarding.

Japan means a great deal to me.
That recognition felt like a real gift.

I believe in the domino effect.
One thing leads to another.
And here we are, having this conversation.

I have been lucky to visit Japan three times, both for work and for personal reasons.
Every time I am there, I have the same feeling: that I am traveling into the future.
I hope that never changes.

What strikes me most as an outsider is the coexistence of two things that rarely live together so naturally.
A deep sense of inner peace rooted in something ancient, and an equally strong pull toward the new, the innovative, and the unexpected.
That duality is rare.

In terms of opportunity, I see real potential in helping organizations visualize concepts with AI.
Precise, meticulous work that reflects how Japanese creators and companies actually think about their products.

Whether that means streamlining production pipelines in film, games, design, architecture, digital art, fashion, and animation studios, or developing creative concepts from the ground up, I think the combination of that cultural precision with an offbeat creative approach is where something genuinely interesting can happen.
I am looking forward to building more connections in Japan.

aratama 璞

Japan has a cultural foundation that not only preserves the past, but also allows sensibilities from different eras to overlap.
If AI is connected not simply to efficiency, but to Japan’s distinctive aesthetics, narratives, sound, space, and physicality, forms of expression may emerge that differ from those developed elsewhere.

The title Hazama (狭間) itself seems to resonate with the Japanese sensibility toward margins, thresholds, and the space between things.
Perhaps there is a quiet connection in making that in-between space the subject of the work.

From what I have seen, corporate use cases in Japan have tended to prioritize cost efficiency and shorter delivery times.
Meanwhile, longer-term experimentation in artistic expression and content development is often carried out by independent creators using their own resources or participating in creative partner programs offered by AI tool providers.

AI competitions are also increasing in number.
But in some cases, there is still room to improve entry fees, judging criteria, and systems for sustained development.
We need structures that reduce the burden on individual creators and connect experimentation to future production opportunities.

Chapter 7　Beyond Film — Live Performance, Music, Code, and Real-Time Expression

Recognition for AI-generated moving-image work currently tends to focus on cinematic short films.
But film production is not the only field being expanded by AI.

Live performance, visuals synchronized with music, advertising, animation, interactive media, programming-based visuals, generative art, and spatial experience design.
New forms are beginning to emerge that cannot be fully understood through the language of conventional filmmaking alone.

Victor Moreno

The potential is absolutely there.
My entry into AI through ComfyUI actually came from TouchDesigner.

TouchDesigner is a node-based visual programming environment for building interactive 2D and 3D applications, real-time multimedia content, and generative art.
The possibilities of integrating AI into structures, textures, and living systems are enormous.
Even in 2026, I think we are using only a small fraction of what is actually there.

I have been building my own style workflows.
By adapting them to the needs of a brief, I can create work that feels more characterful, more on-brand, and more distinctive.

I am an old-school guy who loves The Designers Republic, Tomato, Ryuichi Sakamoto, Bill Viola, Refik Anadol, Ryoji Ikeda, and many more.

Going forward, I want to experiment with integrating TouchDesigner into ComfyUI.
I think it has artistic and abstract potential beyond a purely commercial remit.

aratama 璞

At present, recognition for AI-generated moving-image work still tends to favor cinematic short films.
Many participants also appear to be entering the field through competitions, commissioned work, and the pursuit of social-media reach.

But film production is not the only field being transformed by AI.
Visuals synchronized with music, real-time generation, live staging, audience interaction, and the design of spatial experiences are beginning to produce forms of expression that cannot be measured solely within the conventional category of moving-image work.
The broader value of multidisciplinary and experimental approaches is only beginning to emerge.

Drawing on my earlier background in music, I have been exploring AI audiovisual practice not simply as the screening of a finished video, but as the creation of a resonant space in which music, moving images, the venue, and audience response interact.
Keeping up with emerging technologies and conducting multifaceted experiments will likely remain a long-term process, but I continue to explore whether something genuinely new can emerge from it.
Through AI Creators, I also want to look beyond film and moving images alone, and engage with the broader range of creative expression being expanded by AI.

Chapter 8　A Future Where the “AI” Prefix Disappears

Looking three to five years ahead, Victor Moreno envisions a future in which the “AI” prefix gradually fades away.

Victor Moreno

I hope — and want to believe — that the AI prefix will eventually vanish, absorbed into the backbone of creative pipelines the way other milestones were before it.

AI has blurred the lines between creativity and production.
You do not get a fully polished result from the start, but you can now execute as you create.
Decisions also have to be made faster.

Concept development and prototyping become faster and cheaper.
Brands and clients can visualize where their money is going for a fraction of the investment.
I see AI as a restructuring of how things are made, rather than a disruption.

To build a healthy environment, teams need to become more flexible and project-specific.
Sometimes a creator is the producer, and no production team is needed.
Other times, you need a full structure, with producers and post-producers.

Many brands are pouring money into data specialists.
That makes sense.
But I genuinely hope they come to see the same value in an in-house creative foundation built around AI specialists — people who can drive product prototyping, design, content production, marketing, and storytelling.

And I hope audiences continue to accept AI not as some alien force taking control, but as just another tool — like the typewriter, the camera, the lens, or the pencil.

aratama 璞

I do not think the era in which using AI is valuable in itself will last forever.
What will matter is not which tools were used, but what sensations were created, what people and disciplines were connected, and what remains after the work is experienced.

For AI-native expression to develop, producing works is not enough.
We need to evaluate them, give them context, bring them to the right audiences, and create an environment in which talented creators can continue to work sustainably.

Rather than simply increasing the volume of content, we need systems that discover, present, develop, and connect outstanding creators and works to future production opportunities.
We also need mentors who understand AI-native practice and can offer meaningful evaluation.
Only then will works emerge that audiences remember, discuss, and recognize as the beginning of a new form of expression.

By the time today’s children come of age, legal frameworks and social norms may be more firmly established.
AI creativity may no longer be treated as a special category, but as an ordinary part of the creative environment.

Victor Moreno’s Next Steps — Commercial Production, Narrative Work, and Artistic Expression

Victor Moreno identifies three areas that he wants to continue developing.

Commercial production
Narrative work
Artistic expression

Victor Moreno

I am excited to take on more commercial work, turning briefs into finished products.

At the same time, I have already started working on my next short film.
The international festival recognition for Hazama (狭間) is genuinely encouraging, and I hope it helps open doors to funding as well.

And ultimately, I want to explore the artistic side more freely.
Video art, and visual work synchronized with music and sound.
I believe there is enormous potential beyond the boundaries of commercial requirements.

Editor’s Note｜Building the Next Production Environment for AI Creativity

The value of AI creativity cannot be measured solely by cinematic polish.
AI is expanding creative expression across planning, screenwriting, moving images, music, live performance, illustration, programming, advertising, character development, interactive media, spatial design, and many other fields.

At the same time, raising the quality of the work requires more than asking one AI creator to carry every responsibility alone.
We need to connect individual strengths with specialists, evaluators, technology, funding, time, and opportunities for presentation.

AI Creators aims not simply to introduce people who can use AI tools, but to discover creators and works that open up new forms of expression, and to build an international community in which that talent can continue to develop.
We also want to look beyond cinematic AI filmmaking alone, engaging with emerging formats across live visuals, music, art, code, the Web, illustration, advertising, IP, planning, screenwriting, collaboration, content development, and experimental practice.

Models will continue to improve.
But for a work to stay with an audience, it needs intention, judgment, context, and experience.

What lies beyond AI film is not merely technological progress.
It may be the beginning of a new culture — one that redesigns the creative process, the structure of teams, the criteria by which work is evaluated, and the relationship between creators and audiences.

Turning AI Creativity into Production Opportunities

AI Creators supports the development, promotion, and production of creators pioneering new forms of expression through AI.
We are also building collaborations with companies and organizations exploring AI-driven planning and content production.

Whether you want to learn new forms of expression and refine your work, or develop a project in collaboration with AI creators, explore the pathway that best matches your goals.

en.ai-creators.tech

Generative AI Courses & AI Creator Network

https://en.ai-creators.tech/personal/

Learn practical generative AI skills, build your portfolio, or apply to join our global AI creator network and connect with new creative opportunities.

en.ai-creators.tech

Generative AI Creative Services for Business

https://en.ai-creators.tech/enterprise/

AI Creators supports companies with generative AI creative production, AI creator matching, corporate training, strategic advisory, PR, and partnerships.

The post International AI Creator Dialogue: Victor Moreno × aratama 璞 — AI-Native Expression and the Future of Production first appeared on AI Creators.

AI Animation Is Becoming Infrastructure: Amazon, Netflix, and Chinese Video AI

aratama 璞 | Editor-in-Chief, AI Creators — Thu, 28 May 2026 15:58:34 +0000

In 2026, generative AI in animation and video production has moved well beyond the “useful production aid” stage. It is now transitioning into a production infrastructure — one driven by major streaming platforms, studios, and cloud infrastructure companies.

The moves by Amazon MGM Studios, AWS, Netflix, and Chinese AI video models are not simply more examples of AI adoption. They represent a structural transformation of the entertainment industry — a fundamental redesign of how content is conceived, produced, rights-managed, distributed, and monetized.

This article breaks down Amazon’s and Netflix’s generative AI strategies, the rise of Chinese video AI models, the lessons from the AI dubbing backlash, and four concrete things Japanese companies, production studios, and AI creators should be testing right now.

The Bottom Line

The competition in AI animation is no longer purely about which model renders the most beautiful footage. What matters now is how you embed generative AI across the entire production system — and how you manage quality, rights, and reproducibility at scale.

Why Amazon and Netflix Are Moving Now — AI Animation Shifts from Experiment to Industrial Strategy

To be direct: Amazon and Netflix are no longer treating AI as a purely exploratory technology. Both companies have moved to build organizational structures where AI sits at the center of their production pipelines. Reading these as isolated news stories misses the point. What’s happening is the convergence of four strategic levers — streaming, cloud, production, and IP — into a single, integrated play.

Until recently, generative AI video was largely the domain of independent creators and small studios experimenting with emerging tools. The entry of Prime Video and Netflix changes the category. Generative AI is no longer something production teams “try out” — it is becoming something that fundamentally changes how they operate.

From AI Creators’ perspective, this matters enormously. AI video production cannot be reduced to prompt engineering or tool selection. It only becomes a viable production practice when it spans the full workflow: concept, IP, reference materials, voice, music, editing, rights clearance, and post-release performance analysis. Amazon and Netflix are pointing exactly in that direction.

What Is Amazon MGM Studios and AWS’s “GenAI Creators’ Fund”?

Amazon MGM Studios and AWS have announced the “GenAI Creators’ Fund” — a generative AI animation initiative under which three animated series for Prime Video have been greenlit. The projects are “Cupcake & Friends,” “Love, Diana Music Hunters,” and “Punky Duck.”

According to reports, “Cupcake & Friends” is being produced by BuzzFeed Studios, “Love, Diana Music Hunters” is from creator Albie Hecht, and “Punky Duck” is from Jorge R. Gutierrez, the director behind “The Book of Life.” What’s significant here is not just that Amazon is making AI-assisted animation — it’s that the company is offering a model that bundles funding with a full production environment.

This initiative is designed not merely as tool access but as an end-to-end system: production, distribution, and cloud infrastructure delivered together. It signals that generative AI in video is no longer a single-output problem — it is a workflow architecture problem.
Source: Amazon MGM Studios, AWS Launch the GenAI Creators Fund｜TV Technology

Project Nara Is Not an AI Video Model — It’s a Production Workspace

The technical backbone of the GenAI Creators’ Fund is “Project Nara.” It is not an AI video model in the conventional sense. Reports describe it as a production workspace — an environment that connects creators, AI models, production processes, and cloud infrastructure into a single managed system.

According to coverage, Project Nara is built on AWS and integrates with professional tools already in use across the industry: Maya, Blender, Nuke, Unreal Engine, and Adobe Creative Cloud. In other words, the intent is not to replace existing production tools with AI — it is to connect them through AI and manage the entire pipeline from within that environment.

Some reports indicate that Project Nara is designed to combine multiple external AI models with proprietary models. Given that descriptions vary across sources, this article characterizes it as an integrated workspace that orchestrates external models and production tools rather than specifying individual model names.

For AI creators, the strategic takeaway is clear: “which model is best” is no longer the only question that matters. How you design the full production flow — and who makes decisions at each stage — is becoming the real competitive axis.
Source: Amazon MGM Studios, AWS Launch the GenAI Creators Fund｜TV Technology

Netflix’s INKubator / INK and What “Institutionalizing AI Production” Really Means

A note of caution on Netflix: the company has not issued a formal, large-scale public announcement. However, multiple media outlets have reported — based on LinkedIn profiles and job listings — that Netflix has been quietly building a generative AI animation team known as “INKubator” or “INK.”

Reports suggest INKubator is structured as a “GenAI-native” animation studio, starting with short-form animated content and specials, with longer-form ambitions potentially to follow. Serrena Iyer is reported to be playing a central role. This should be treated as information drawn from job postings and industry reporting — not a formal Netflix press release.

What matters strategically is the implication: Netflix appears to be integrating generative AI not just into post-production efficiency, but into the upstream stages of development — short-form creation, concept validation, and world-building. AI production is beginning to look less like an experiment and more like an institutionalized practice.
Source: Netflix is building an AI animation studio｜The Verge

What the Streamers Are Really After — Not Content, but the “Production OS”

If you read Amazon’s and Netflix’s moves simply as “big streamers making AI animation,” you’re missing the deeper play. What both companies are positioning for is not content — it’s a layer above content: the Production OS of the entertainment industry.

The Production OS, as we use the term here, refers to the full stack of rules, tools, cloud infrastructure, rights management, quality control, and distribution that underlies how content gets made. When generative AI enters the picture, a single key visual, a character design, a voice recording, or a reference clip becomes connected to multiple AI generation stages across the pipeline.

This creates an entirely new design challenge: which assets go in, which AI models handle them, where do humans review and revise, and where does final approval happen? The streamers are not just producing AI content. They are racing to define the standard for how AI-era production flows are structured.

Amazon’s Play: Connecting AI Content to AWS Production Infrastructure

Amazon’s ambition extends well beyond Prime Video’s animation slate. By building a production environment like Project Nara on AWS, Amazon is creating a pathway where content creation and cloud infrastructure converge — a structure in which production workflows run on Amazon’s own systems.

Video production is becoming increasingly data-intensive. High-resolution imagery, multi-track audio, 3D assets, motion data, editorial timelines, rights metadata — the volume of production data keeps growing. Add generative AI, and you layer on top of that: model processing, version control, prompt histories, and reference asset management.

Amazon’s move into AI animation can be read as a content play — but also as a move to pull professional creative workflows into the AWS ecosystem. The entity that controls the production infrastructure gains meaningful influence over how production costs are structured and how workflows evolve.
Source: Amazon MGM Studios and AWS Launch GenAI Creators Fund｜Variety

Netflix’s Play: AI Integration Beyond Post-Production, Into Development

To understand Netflix’s strategy fully, you need to consider the InterPositive acquisition alongside the INKubator reports. In March 2026, Netflix officially announced the acquisition of InterPositive, a film production technology company co-founded by Ben Affleck. The purchase price was not disclosed publicly.

According to Netflix’s official statement, InterPositive develops AI-assisted tools built for filmmakers — systems trained on actual production footage to support editing consistency, lighting correction, missing shot completion, and visual coherence across a production.

This move signals that Netflix sees AI not as a video generation shortcut, but as a technology that supports the entire production process. If InterPositive handles post-production and production support, and INKubator (per reporting) addresses short-form creation and concept validation, then Netflix appears to be assembling capabilities that span from upstream development all the way through to delivery.

Netflix has also published a generative AI usage guideline for production partners. It specifies that intended GenAI use should be disclosed to the relevant Netflix team, and that cases involving personal data, talent likeness or voice, third-party IP, or final deliverables require careful review and approval.

This is not a company treating AI as a sandbox. Netflix is beginning to manage AI as a controlled production practice.
Source: Why InterPositive Is Joining Netflix｜Netflix
Source: Using Generative AI in Content Production｜Netflix Partner Help Center

What “Human-Led, AI-Supported” Actually Means

When Amazon and Netflix discuss AI, both companies consistently use language that frames human creativity as central, with AI in a supporting role. This reflects genuine respect for creators and audiences — but it is also a carefully calibrated risk-management position.

The 2023 WGA and SAG-AFTRA strikes made generative AI a central labor issue in Hollywood. WGA secured AI-related protections in its 2023 contract, and SAG-AFTRA has published AI-specific resources covering how actors’ likenesses, voices, and performances may be used.

“Human-led” is not a PR phrase. It is a design principle — one that determines where in the production process human judgment is required. The question is no longer whether to use AI. The question is which workflows AI may enter, with whose consent, to what quality standard, and for what purpose.
Source: Artificial Intelligence｜Writers Guild of America West
Source: Artificial Intelligence Resources｜SAG-AFTRA

The Rise of Chinese AI Video Models — How Seedance, Kling, and Hailuo Are Reshaping Production

While Amazon and Netflix dominate the headlines, another force is reshaping entertainment production at the ground level: Chinese AI video models. ByteDance’s Seedance, Kuaishou’s Kling, and MiniMax’s Hailuo have become significant players in the field — advancing rapidly across video generation, character consistency, reference-based control, and audio synchronization.

What distinguishes Chinese AI video models is not just visual spectacle. It’s that they are closing the gap with actual creator workflows. We’ve moved past “text-to-video as a novelty.” These models now accept combinations of images, video clips, audio, character references, and motion references as inputs — which is exactly how professional production works.

That said, high capability and commercial safety are separate issues. Anyone using Chinese AI tools in a commercial context needs to scrutinize terms of service, training data practices, output ownership, data handling, and the evolving landscape of international regulatory risk.

Seedance 2.0: Beyond Video, Into Audio, Reference Control, and Editorial Direction

ByteDance’s Seedance 2.0 is a strong indicator of where AI video generation is heading. The ByteDance Seed official page describes Seedance 2.0 as an integrated multimodal audio-video generation architecture supporting text, image, audio, and video input.

The official blog elaborates that the model can process multiple images, video clips, audio tracks, and natural language instructions as simultaneous inputs. This marks a clear evolution: AI video is no longer just about generating footage from a text prompt. It is becoming a production experience that encompasses reference assets, audio, staging direction, and editorial control.

From AI Creators’ perspective, the critical insight here is what this means for the creator’s role. Prompt engineering, while still relevant, is no longer the whole job. The more consequential skill is reference design — deciding which images anchor the visual identity, which audio defines the tone, which clip defines the motion, which elements are fixed and which are delegated to the model. That is where creative direction now lives.
Source: Seedance 2.0｜ByteDance Seed
Source: Official Launch of Seedance 2.0｜ByteDance Seed Blog

Kling: Evaluating It Through Character Consistency and Commercial Viability

Kling deserves to be assessed not as a flashy AI video novelty, but in terms of character consistency and real-world production usability. On AI Creators, our coverage of Kling Motion Control 3.0 has focused on its use of multiple reference inputs to improve facial consistency and motion coherence across shots.

Character consistency is one of the hardest unsolved problems in AI video production. Generating a single beautiful shot is achievable. Maintaining the same character across multiple angles, expressions, and camera moves — without visual drift — is where most models still fall short.

That is precisely why models like Kling attract serious attention. For brand campaigns, character IP, animated series, and advertising, visual reproducibility matters more than one-off aesthetics. Before AI video can enter professional production workflows, it has to demonstrate control, not just capability.

en.ai-creators.tech/media

Kling Motion Control 3.0: Is AI Video Character Consistency Becoming Producti...

https://en.ai-creators.tech/media/creative/kling3-motion/

Character consistency has long been one of the biggest bottlenecks in AI video generation — and Kling Motion Control 3.0 suggests that this may be starting to change. The update marks a meaningful step toward more production-ready control, helping creators better preserve a character’s facial identity and motion coherence across more complex scenes through the use of multiple reference inputs. Instead of relying on repeated prompt attempts and hoping for a usable result, AI video workflows ar...

Commercial Risk When Using Hailuo and Other Chinese AI Tools

The intellectual property risks surrounding Chinese AI video models are not hypothetical. Reuters has reported that Walt Disney, Comcast’s Universal, and Warner Bros. Discovery have filed copyright lawsuits against China’s MiniMax and its AI image and video generation system, Hailuo. In May 2026, Reuters further reported that MiniMax’s motion to dismiss the Disney lawsuit had been denied.

This is not a Hailuo-specific issue. As AI video models become more capable, so does their ability to generate footage resembling existing characters, brand assets, and intellectual property. That versatility is powerful — and commercially dangerous.

Data management risk: You need to understand how input materials, prompts, and outputs are stored and whether they may be used for model training.
IP and copyright risk: You need to verify that generated output does not substantially resemble existing characters, brands, or copyrighted works.
Regulatory and contractual risk: You need to confirm that your use of the tool complies with the platform’s terms of service, client contracts, and applicable local regulations.

Our position at AI Creators is not that Chinese AI tools should be avoided. The position is that before using any of them commercially, you must resolve questions around rights, source materials, contracts, and data retention.
Source: Disney, Universal, Warner Bros Discovery sue China’s MiniMax for copyright infringement｜Reuters
Source: China’s MiniMax loses bid to end Disney copyright lawsuit over AI system｜Reuters

Why Success Stories Aren’t Enough — What the AI Dubbing Controversy Revealed

Looking only at success stories in generative AI production is a mistake. The failures teach more. The AI dubbing backlash is a case study in why quality, consent, transparency, and professional respect are not optional — they are prerequisites for AI content to be socially accepted.

Audiences don’t object to AI simply because AI was used. In most cases, the anger reflects something deeper: a suspicion that human expression was treated as expendable, that voice actors and rights holders weren’t consulted, that the work wasn’t taken seriously. The technology is not the problem. The disrespect that technology can enable — that is.

AI production is a domain where technical feasibility and cultural acceptance do not automatically align. That gap cannot be closed by better models alone.

What Actually Happened with Amazon’s AI Dubbing Problem

Amazon Prime Video faced significant criticism over AI-generated dubbing. GamesRadar reported that several anime titles on Prime Video were found to carry dubs labeled “English [AI beta],” and that some of those AI dubs were subsequently removed. Titles cited in reports include “Banana Fish,” “No Game, No Life Zero,” and “Pet.”

The problem was not purely about output quality. According to reporting, rights holders and studios associated with some of the affected titles indicated they had not been informed of or approved the use of AI dubbing — a claim that extended concern well beyond the viewing audience into the professional community.

Dubbing touches voice performance, emotional delivery, and the audience’s relationship with characters. When AI dubbing misses the mark — whether in technical quality or tonal fidelity — viewers do not experience it as a neutral efficiency measure. They experience it as the work being treated as less than it deserves.

For any organization considering AI dubbing or AI voice applications, quality review is not sufficient on its own. Consent, clear labeling, and alignment with rights holders and production partners are essential — particularly for established titles and beloved IP, where fan expectations are unforgiving.
Source: After more AI English anime dubs were removed from Prime Video｜GamesRadar+

Three Conditions Required to Avoid Being Called “AI Slop”

In international discourse, “AI slop” has become shorthand for low-quality, mass-produced AI content that feels disposable and disrespectful of the medium. Avoiding that label requires meeting three conditions — none of which are purely technical.

Quality: The footage must be structurally coherent. Voice and performance must feel intentional. The editorial rhythm must work as content — not as a test render.
Consent: The rights status of every input element — people, voices, characters, music, branded assets — must be confirmed before production begins.
Purpose: AI must be used for a reason that goes beyond cost reduction — expanding creative range, accelerating prototyping, unlocking production options that weren’t previously viable. The intent must be legible.

The verdict on AI use is never simply “you used AI.” It is always which workflow, with whose consent, to what standard, and to what end. That is a question of production ethics and decision architecture — not tooling.

en.ai-creators.tech/media

AI Creative Direction: How to Rise Above AI Slop

https://en.ai-creators.tech/media/creative/ai-slop/

Generative AI has made it possible for almost anyone to produce convincing visuals, copy, and video in a fraction of the time it once took. The tools can give you something that looks finished. But those same tools have also flooded the market with output that reads, at a glance, as unmistakably AI-generated — generic in composition, flat in tone, and empty of context.The question being asked now isn't whether you're using AI. It's how deeply you've thought, designed, and ...

How the AI Creator’s Role Is Changing — From Prompt Craftsperson to Production Architect

This is where AI Creators’ editorial position matters most. Generative AI’s evolution is unquestionably changing what it means to be an AI creator. But that change is not simply “some jobs will disappear.” The more accurate framing is this: the value of people who can determine what to create with AI, design how to create it, and decide where human judgment is required — that value is rising.

AI creators have often been characterized as people who excel at prompting — generating high-quality images or video through careful input engineering. In commercial production contexts, that is a starting point, not a complete skill set. What clients and studios increasingly need is someone who can design the entire production.

AI Creators’ position is that the AI creator’s role is shifting from “generator” to production architect. Knowing the tools is the baseline. What differentiates is the ability to design intent, structure, operations, and risk management across the full production lifecycle.

What’s Needed Now: Not “Can Generate” but “Can Design Production”

The most common misconception in AI video production is that generating one excellent shot is enough to build a practice. Output quality matters — but in commercial contexts, the ability to produce the right footage reliably and repeatedly is worth more than a single stunning result.

Consider a brand promotional video featuring a proprietary character. A single strong shot is not a deliverable. The character needs to hold up from multiple angles, with changing expressions, in motion, synchronized with audio, maintaining brand tone across every format — social, vertical short, web, event screen.

That requires more than prompts. It requires character documentation, reference asset design, shot structure, editorial guidelines, and revision protocols. The AI creator who is ready for commercial work is not the one who operates generation tools. It is the one who understands the production objective, determines where AI belongs in the workflow, and ensures the outcome is delivery-ready.

Three Capabilities Clients and Studios Are Looking For

In AI production practice, the capabilities that matter most resolve into three areas.

Creative Direction: The ability to define what needs to be created and to evaluate AI output against a clear creative intent. This means designing the content’s purpose before writing a single prompt.
Production System Design: The ability to build workflows that are repeatable. Not a beautiful one-off image, but a pipeline that can produce series-ready, scalable content.
Risk and Rights Management: The ability to integrate asset clearance, output rights, voice and music licensing, and reputational risk assessment into the planning stage — not as an afterthought.

As Netflix’s generative AI guidelines make clear, commercial production with GenAI requires that intended use be shared with stakeholders and that risk be actively evaluated. Creative excellence and accountability are not competing values — in this environment, they are inseparable.
Source: Using Generative AI in Content Production｜Netflix Partner Help Center

AI Expands Perception and Editorial Judgment — It Doesn’t Replace Them

The real nature of AI animation is not the substitution of hand-drawn craft. It is something more interesting: the rapid visualization of possibilities that a human creator could not have surveyed manually — followed by a human choosing which of those possibilities carries meaning.

Generative AI tools produce enormous volumes of “almost-something.” Most of it won’t become content. But somewhere in that output, occasionally, there’s a frame, a motion, a composition that makes you stop and say — that’s it. And the ability to recognize that moment is not a function of the model. It is a function of the person watching.

What AI-era creators need, more than generation skill, is the ability to identify “what could become work” from an overwhelming volume of machine output — and to embed that editorial judgment into a repeatable production system. That is not prompt craftsmanship. It is the perspective of an editor or director. And it cannot be automated.

What Companies, Studios, and AI Creators in Japan Should Verify Right Now

For Japanese companies, production studios, and AI creators, the right first move is not launching an ambitious AI animation project. It is running small, deliberate validations to identify what a viable AI production flow actually looks like for your specific context.

Generative AI is not a plug-in solution that immediately increases production efficiency. Without preparation, common failure modes include inconsistent output, unresolved rights questions, no viable revision path, and workflows that only one person can reproduce.

If your organization is beginning to explore AI video production, start with a short-form proof of concept — something in the 15-to-60-second range. Even at that scale, running the full loop — concept, reference design, generation, editing, audio, quality review, publication, performance analysis — will surface exactly where your gaps are.

Validation 1: Run a Full Production Loop with a 15- to 60-Second PoC

Don’t start with a feature-length ambition. Begin with a short-form proof of concept and complete every stage of the production process: concept development, storyboarding, reference asset selection, generation, editing, audio, subtitles, publication, and post-release analysis.

Even a 15-second brand clip will expose the issues that matter: Can you maintain consistent character appearance? Does camera movement land as intended? Does audio sync with footage? Can you act on revision notes? Are there any rights concerns in the input materials?

The goal of a short-form PoC is not to produce a polished deliverable — though that matters too. The goal is to identify which stages consume the most time, which stages require human judgment, and which elements can be templatized and reused across future projects.

Validation 2: Run a Character Consistency Test with Your Own IP

For organizations with proprietary characters or brand IPs, one of the most important early tests is consistency across shots. Can you maintain facial features, costume, voice, world design, and camera style across multiple cuts?

For character-driven IP — whether animation, brand mascots, or game characters — what matters is not a single beautiful frame but series viability. The same character needs to hold up across social posts, promotional video, vertical format, web content, and live event material. If the character drifts between executions, the brand breaks down.

If you’re using AI video models with reference input capability — such as Kling or Seedance 2.0 — don’t rely on the model to make these decisions. Define your own rules: which visual elements are non-negotiable, and which can be interpreted freely by the system.

Validation 3: Establish Rights and Asset Management Rules Before Production Begins

In commercial production, rights management is not a post-production task — it is a pre-production requirement. Before a single frame is generated, you need clear answers: whose image is being used, whose voice, which music, which reference clips, and whether you have the rights to use all of them in the way you intend.

The elements that require the most scrutiny are people, voices, existing characters, branded assets, and music. AI tools generate new output from input materials — but having the output does not mean you had the rights to use the inputs. In addition, some platforms may retain or train on uploaded materials; that question should be resolved at the contract stage, not after delivery.

A 15-minute rights review before production starts can prevent months of legal exposure after it ends. In AI production, preparation before generation is what determines quality and legal safety.

Validation 4: Build an Editorial Layer — Never Deliver Raw AI Output

AI-generated footage is source material. It is not a finished deliverable. Treating it as one is among the most common and costly mistakes in AI production today. Raw output frequently contains visual inconsistencies, audio misalignments, subtitle errors, brand tone drift, and unresolved rights questions — none of which a model will flag for you.

Making AI output commercially viable requires a dedicated human editorial layer: video editing, audio treatment, color grading, subtitling, narration, rights confirmation, and quality sign-off. These are not optional. They are the production steps that transform AI output into content that can be published and stand behind.

A budget allocation that invests heavily in generation and cuts corners on finishing is a budget that compromises both quality and trust. AI-era production teams need a defined collaboration structure between generation leads, editors, directors, and rights reviewers.

Conclusion — The AI Animation Competition Has Moved from Model Performance to Production System Design

Looking across Amazon’s, Netflix’s, and China’s AI video model developments, a common structural pattern emerges. The competition in AI animation is no longer decided by which model renders the most visually striking footage. It is decided by how you integrate generative AI into a production system that encompasses concept, execution, rights management, distribution, and monetization.

Amazon MGM Studios and AWS’s GenAI Creators’ Fund, Project Nara, Netflix’s acquisition of InterPositive, the INKubator reports, Netflix’s generative AI production guidelines, ByteDance’s Seedance 2.0, Kling’s character consistency advances, and the copyright litigation around Hailuo — these are not unrelated developments. They share a single underlying story: generative AI is beginning to restructure the production architecture of the entertainment industry itself.

Which of these approaches will prevail? That remains genuinely uncertain. What is no longer uncertain is this: the gap between organizations that use AI occasionally and organizations that have embedded AI into their production design is becoming harder to close.

For Japanese companies — and any studio navigating the same AI production shift — the answer is not to chase every new tool. It is to develop the AI-native production design capability that fits your IP, your workflows, and your team. Not by copying someone else’s success story — but by testing your own flow, making your own judgments, and building your own system. That iterative practice is the only durable competitive advantage available now.

The post AI Animation Is Becoming Infrastructure: Amazon, Netflix, and Chinese Video AI first appeared on AI Creators.

AI Creative Direction: How to Rise Above AI Slop

aratama 璞 | Editor-in-Chief, AI Creators — Sun, 05 Apr 2026 03:45:41 +0000

Generative AI has made it possible for almost anyone to produce convincing visuals, copy, and video in a fraction of the time it once took. The tools can give you something that looks finished. But those same tools have also flooded the market with output that reads, at a glance, as unmistakably AI-generated — generic in composition, flat in tone, and empty of context.
The question being asked now isn’t whether you’re using AI. It’s how deeply you’ve thought, designed, and refined what comes out of it.

The market has already moved past its first phase. What clients are increasingly asking for isn’t “no AI” — it’s “AI is fine, but it shouldn’t look like AI.”
Audiences are no longer impressed by the fact of generation. The bar has shifted: what matters now is whether the output can survive revision, hold up across formats, and serve the brand over time. That means the gap between strong and weak work is no longer about output volume. It’s about the quality of the design decisions and editorial judgment behind it.
This article uses the concept of “AI slop” as a lens to diagnose the problem — and then lays out a framework for the kind of creative direction that commands premium-rate client work.

Chapter 1 | What Is AI Slop, and Why Does It Matter Now

“AI slop” is a term that emerged in English-speaking creator communities to describe high-volume, low-intention AI output — work that is generic, context-free, and interchangeable.
It isn’t a criticism of AI-assisted work in general. The problem isn’t the technology. It’s the uncritical, unedited output that gets passed off as finished work.

AI slop tends to share a recognizable set of traits. The compositions feel familiar. The copy is flat. There’s no sense of brand or context. The textures are either slightly off or so polished they feel sterile. It might look acceptable in the moment, but there’s nothing that stays with you. That’s the pattern.

Why is it a problem now, specifically? Because the volume has crossed a threshold. Novelty — “look what the tool can do” — no longer carries weight on its own. Work is now being evaluated on quality, intent, and meaning.

The economic consequence is structural. When everything looks like it could have been made by anyone, it creates a market-wide assumption that AI work should be cheap. That’s what drives rate compression.

From a client’s perspective, the damage runs deeper than aesthetics. AI slop is structurally difficult to work with: when revision requests come in, there’s no rationale to explain why the creative went in that direction. Brand tone shifts with each new execution. Consistency erodes over time. Internal sign-off becomes harder to secure. This kind of operational breakdown quietly undermines trust — and it compounds. The intention was to cut costs. The result is damage to brand equity. That’s the real risk of AI slop.

Chapter 2 | AI Slop Comes from Outsourcing Your Judgment, Not from the Tools

Japanese news segment (TBS Cross DIG with Bloomberg) examining the productivity risks of uncritical AI use in professional workflows

Blaming the tools for AI slop misses the point. The same models, the same platforms — the output quality varies dramatically depending on who’s using them and how. The problem isn’t on the tool side. It’s in the depth of thinking on the human side.

The issue is treating the first thing that comes out of a prompt as a finished deliverable. It isn’t. Generated output is raw material. Think of it like ingredients arriving at a kitchen. What happens next — what you keep, what you cut, how you shape it — that’s where the actual work begins.

“The AI is good at this” and “you are good at this” are not the same statement. AI returns probabilistic output in response to instructions. Giving that output intention, context, and quality assurance is a human responsibility. When that distinction gets blurred, real problems follow in production:

The work arrives finished but purposeless — it’s unclear what it was supposed to communicate
Quality is inconsistent and can’t be reliably reproduced
Revision requests and intent questions can’t be answered with any real rationale
Clients don’t feel confident commissioning follow-on work

Hands-off prompting can look efficient. But it produces work no one wants to stand behind — and over time, it’s a production style that doesn’t build trust. The distinction isn’t just about output quality. It’s about who takes responsibility for what the work says and does. Shifting away from that hands-off posture is the starting point for attracting higher-value client work.

Chapter 3 | What Creative Direction Actually Means for Premium-Rate Work

So what separates directed work from AI slop? The core isn’t about polishing output. It’s about designing the intent and context before and during the process — that’s creative direction. Here are six principles that define it.

3-1. Design the Intent

Before production begins, articulate who the work is for, what it should communicate, and what it should make them feel.
A campaign built to grow awareness and a campaign built to drive direct purchase have entirely different definitions of “right” — and entirely different creative answers, even when the visual quality looks equivalent. Even when clients haven’t clearly articulated those goals themselves, helping them define the brief is part of the director’s value.

3-2. Treat Generated Output as Material, Not Result

Not accepting the first output as the final output is the foundation of quality.
Rather than publishing the first image that comes out, the practice is to generate multiple options, compare them, and make deliberate decisions about what to cut. What distinguishes strong work isn’t the volume generated — it’s the precision of the selection and editing process. The craft lives in what you choose not to use.

3-3. Maintain a Consistent Brand World

Color, tone, composition, emotional register, sense of light, warmth of language — keeping these aligned is what makes a body of work feel coherent.
When the tone shifts between a banner, a social post, and a landing page, brand credibility takes a visible hit — and that inconsistency becomes expensive to correct late in production.
A single striking execution matters less than a body of work that reads as one continuous experience — that’s what builds brand equity. Visual and tonal consistency is also, practically, one of the most effective ways to prevent work from reading as AI-generated.

3-4. Catch Incongruities and Fix Them

In images, it might be the hands or the eyeline. In copy, it might be a tonal inconsistency or an off-register word — something that reads fine in isolation but breaks the voice of the piece. These small things have an outsized effect on how the whole piece lands.
People respond to these details even when they can’t explain why something feels wrong. Not ignoring “something’s off” is what determines final quality. Having that sensitivity — and acting on it — is something tools can’t replicate.

3-5. Inject Context

Brand tone, industry context, audience state of mind — these need to be embedded in the work, not assumed.
The same idea of “warmth” looks completely different in a luxury brand context versus a public sector communication. The final step is overwriting the generic with something specific — giving the work a voice that belongs to someone. That’s how you move past output that looks competent but belongs to no one.

3-6. Build for Quality Control and Reproducibility

A single strong execution matters less, in practice, than being able to produce at the same standard the next time.
The work needs to hold up under revision, across formats, and throughout an ongoing relationship.
At the higher end of client work, what’s being evaluated isn’t just the final deliverable — it’s whether the production system behind it can be trusted. “I know this person will deliver the same standard again” is what generates repeat commissions.

Chapter 4 | What the Difference Looks Like in Practice

The same brief, handled two different ways, produces work that feels fundamentally different.
Take a concrete example: social media content for a sustainable apparel brand, targeting urban consumers in their late twenties to early thirties.

Output without Direction

Generic bright-natural-light product imagery used as-is
Copy along the lines of “Make the eco-friendly choice”
A string of hashtags with no editorial logic
No clear sense of who it’s for
No identifiable brand personality
Nothing that stays with you the next day

Creatively Directed Output

Color temperature and texture aligned to the brand’s visual identity
Copy that speaks in the audience’s own register
Deliberate selection — what’s included and what’s left out
The full post reads as a single, coherent experience
Details that give you something to look at twice
Designed to hold up under revision and across formats

The undirected output is missing two things: a judgment about audience, and a deliberate decision about what to keep.
The directed output goes further — it’s designed not just for how it looks, but for how it will be read. That difference determines how the finished work is received.

But the value difference isn’t only visible in the initial output. It shows up when the client asks for a revision, when the campaign needs to extend to a new format, or when the same brief has to be executed six months later. Directed work is built to survive those conditions. Undirected work rarely is. When the revision conversation comes, or the brief for the next execution, and the response inspires confidence — that’s what compounds into higher-value work.

Chapter 5 | Why Creative Direction Is What Commands Premium Rates

“I can use AI” is no longer a differentiator. Using it is assumed. What’s being evaluated is what you do with it.

Premium rates aren’t paid for visual impact. They’re paid for the confidence of handing over a decision. More specifically, they’re paid for the confidence that someone can make and defend creative decisions under real production conditions — through briefs that shift, revisions that multiply, and stakeholders who need things explained.

Clients are paying not just for the final deliverable, but for the soundness of the design rationale, the ease of revision, and the ability to explain creative choices clearly.
Someone who can be trusted across the full scope — from intent design to quality control — is genuinely hard to replace. Being commissioned by name, specifically, comes from having a point of view, a set of standards, and a design process — not just from making attractive work.

From a client’s perspective, there are four conditions that make it feel safe to commit real budget to a creative partner:

Reproducibility — consistent quality across executions
Accountability — the ability to explain why the creative went where it did
Brand alignment — genuine understanding and control of the brand’s visual and tonal identity
Operational confidence — reliable handling of revisions, pivots, and format extensions

Premium rates aren’t simply a pricing question. They’re the return on trust.
And that trust is built not through generation speed, but through the depth of judgment, design, and editing.

Chapter 6 | The Skill That Matters Now Is Editorial Judgment, Not Generation Speed

Generative AI has dramatically lowered the barrier to making things. Creative work that once required specialized skill can now be produced in rough form by almost anyone, in a short time. That’s a genuine and significant shift.

But as a consequence, value has migrated — away from output volume, toward the ability to shape output into something meaningful.
If anyone can produce a version of something, the act of producing it stops being rare. What remains rare is the capacity to give that output intention, context, and coherence — to catch what’s wrong with it, to be accountable for what it says.

Conceiving, selecting, unifying, correcting, explaining, finishing — taken together, that’s what creative direction is. It’s not a position against AI. It’s the editorial and decision-making work that makes AI output worth something.

What we consider valuable isn’t the ability to generate — it’s the ability to shape output into something that means something.
The priority isn’t using AI fast. It’s making considered, accountable creative decisions. Not whether to use AI, but how to design and refine what it produces. That’s the position AI Creators is built on.

Chapter 7 | How to Position Yourself to Attract Higher-Value Work

Given all of this, how do you make the value visible? Leading with “I can use AI” isn’t enough. You need to position yourself as someone who can direct AI toward a specific purpose.

In portfolios and proposals, the finished work is only part of what matters. Three things are worth showing:

The final work — the quality of the output itself
Creative intent — why this direction, for whom, communicating what
The revision and refinement process — before/after, and a record of the decisions made in between

Showing all three moves a portfolio from a collection of work to a demonstration of practice.
The impression shifts from “this person can make things” to “this person can be trusted with a brief.” A portfolio built this way isn’t just proof of output — it’s proof of judgment. And judgment, in a market flooded with generated work, is the thing that actually differentiates.

What matters for your professional brand isn’t the number of pieces — it’s the consistency of your point of view. Having clear standards for what you consider strong work and what you’ll push back on, and communicating that through your writing and your output — that’s how you build a reputation as a director-level creative, not just a producer.

Conclusion | Sinking into AI Slop, or Designing for Value

The problem with AI slop isn’t AI. It’s the habit of generating without judgment and releasing without editing. Now that the technology is everywhere, the distinction will be between people who generate a lot and people who make deliberate, considered decisions about what they produce.

High-value client work follows from creative direction, not generation speed.
Designing the intent, selecting the material, maintaining the brand world, catching what’s wrong, injecting context, managing quality — this full chain of design thinking is what accumulates as trust and returns as value.

Three things worth starting today:

Get into the habit of writing a brief intent document before production starts
Treat generated output as raw material, not a finished deliverable
Add your decision-making process to your portfolio

The market is no longer rewarding generation alone. It is rewarding judgment, coherence, and the willingness to be accountable for what you make.
What becomes valuable from here is the ability to take what gets generated and make it mean something — to stand behind it as professional, accountable creative work.

Is your AI practice ending at generation — or are you designing all the way through to value?

The post AI Creative Direction: How to Rise Above AI Slop first appeared on AI Creators.

What Is a Generative AI Artist? Skills, Monetization & Roadmap

aratama 璞 | Editor-in-Chief, AI Creators — Sat, 28 Mar 2026 15:00:34 +0000

Generating high-quality images and videos with AI is no longer a rare skill. However, there is a massive gap between producing compelling work and presenting a unique perspective through a cohesive body of work.

The question today isn’t just whether you can master the tools. What matters is the vision you explore through generative AI, the world you build, and how you connect that work to cultural and contemporary contexts. Those who can shape that entire process are the ones we call “Generative AI Artists.”

A Generative AI Artist is not merely an operator outputting AI files; they are an artist shaping personal inquiries and ideas into a continuous body of work. They are evaluated by different standards from “AI Creators,” who typically focus on practical tasks and problem-solving. For an artist, the focus extends beyond the standalone piece to the underlying context, editorial judgment, transparency, and consistency across their practice.

This article clarifies the definition of a Generative AI Artist, outlining the differences from an AI Creator, the necessary “5 Pillars of Artistry,” monetization structures, copyright and ethical considerations, and a roadmap for progressing from a beginner to the global stage.

A Generative AI Artist is a creative voice presenting unique inquiries and philosophies, not just an operator generating outputs with AI.
The distinction from an AI Creator lies not just in the quality of the deliverables, but in authorship, consistency, contextual grounding, and underlying philosophy.
Their scope of activity extends beyond independent production to exhibitions, brand collaborations, IP development, and educational or research partnerships.

What is a Generative AI Artist?

Defining the Generative AI Artist

AI Creators defines a Generative AI Artist as an “artist who treats generative AI not merely as a tool, but as a medium—elevating it through their own sensibilities, editorial judgment, and technical choices to consistently present a unique worldview.”

What matters is treating AI as a medium, not just a tool. This requires the flexibility to embrace unpredictable deviations and the randomness of AI models as a core part of the creative process.

The Fundamental Difference from “Someone Using AI to Make Images”

With current tools, anyone can output beautiful images in seconds. However, technical proficiency alone does not guarantee artistic value or authorship.
The difference between “someone using AI to make images” and an “artist” lies in the continuity of the work and whether it wrestles with questions like “why this motif?” or “why this texture?”
It is not just about the perfection of a single image, but the underlying theme: what is being observed and why it is being presented to the world.

Why It Is Necessary to Distinguish These Terms Now

AI tools have become mainstream, enabling anyone to produce strong work. Consequently, the distinction of “who is making this, and why” has become more critical than ever.

REFIK ANADOL (B. 1985) Machine Hallucinations – ISS Dreams｜Christie’s Press

In March 2025, Christie’s held the first dedicated AI art sale by a major auction house, concluding with a total of $728,784.
(Source: Christie’s)

Such developments indicate that AI expression is increasingly treated within the context of contemporary art, rather than merely as technical demonstrations. Therefore, there is growing significance in distinguishing between the AI Creator as a practical professional and the Generative AI Artist as an author of a body of work.

Additionally, we cannot overlook the narrative example of NEON ONI, an artist who transitioned from generative AI into a “real band.”
This digital IP, originating from Suno AI, achieved 80,000 monthly listeners on Spotify and advanced to the Japanese qualifiers for Wacken.

en.ai-creators.tech/media

From AI-Generated IP to Live Band: What NEON ONI Teaches Brands

https://en.ai-creators.tech/media/marketing/neon-oni/

The emergence of generative AI has made it possible for small teams to produce high-quality creative output. As a result, many AI creators and corporate marketing professionals now face a central challenge: how to turn these generated assets into sustainable businesses and dedicated fan communities.Within this context, an early case study has emerged that is attracting attention not only in the music industry but also in broader marketing and intellectual property (IP) design. This is the fic...

Differences Between AI Creators and AI Artists

There is no inherent superiority between the two. Because their objectives, evaluation criteria, and time horizons differ, it is important to clarify which direction (or both) you wish to pursue.

【Comparison: AI Creator vs. Generative AI Artist】

Comparison Metric	AI Creator	AI Artist
Primary Objective	Problem-solving, commercial outcomes	Self-expression, cultural inquiry, aesthetic exploration
Starting Point	Requirements, client briefs, target audience	Philosophy, internal motivation
Source of Value	Implementation, reproducibility, speed	Authorship, context, scarcity
Core Evaluation	Quality, delivery time, business results	Concept, worldview, consistency
Primary Fields	Advertising, social media, corporate projects	Exhibitions, brand collaborations, IP, culture
Time Horizons	Project-based	An ongoing body of work
How Companies See Them	Production and implementation partner	Co-creator enhancing brand value

Objective: Problem-Solving vs. Expression

AI Creators aim to solve client problems and generate commercial results. In contrast, Generative AI Artists operate primarily for self-expression, societal inquiry, and the exploration of aesthetics.

Starting Point: Requirements vs. Philosophy

AI Creators work backward from requirements definitions and target audience design. Generative AI Artists construct their work starting from internal motivations, philosophies, and aesthetics.

Evaluation: Deliverables vs. Context

The value of an AI Creator is measured by output quality, reproducibility, and production efficiency. A Generative AI Artist is evaluated on the strength of their concept, the consistency of their creative world, and the context they present.

Time Horizons: Project vs. Worldview

A creator’s work is segmented by project, whereas an artist’s activities are sustained over a long-term series or a lifelong creative practice.

How Companies See Them: Production Talent vs. Brand Co-Creator

For corporations, AI Creators are excellent production and technical partners. Generative AI Artists, however, are often positioned as co-creators who expand the brand’s core values.

The “5 Pillars of Artistry” for Generative AI Artists

Layer 1: Technical Foundation

The first requirement is a technical foundation to operate generative AI. This demands an understanding of the characteristics of major generation models across images, video, and audio, and the ability to apply them strategically.
Specifically, this includes prompt design, workflow construction, and the training and fine-tuning of custom models or LoRAs.

Layer 2: Expressive Foundation

The expressive layer transforms technology into art. It is not enough to simply arrange generated outputs; the focus is on what to show and how to show it.
This requires conceptual design skills, a consistent worldview, an understanding of art history and visual grammar, and the editorial capability to select and curate generated results.

Layer 3: Contextual Design

The ability to design context connects the artwork to society. Beyond the visual impact of a single piece, an artist must consider the underlying issues it addresses and how it reaches others.
This involves the ability to write artist statements, articulate concepts for exhibitions and critique, and understand rights, ethics, and transparency.

Layer 4: Visibility, Connection, and Career Development

Creating art doesn’t automatically generate opportunities. Modern Generative AI Artists must be able to communicate their work in their own words, present it to the right audience, and connect it to exhibitions, collaborations, and commercial projects.
Visibility involves managing social media, building a portfolio, publishing statements, and introducing yourself (often in English). Connection means proactively reaching out to galleries, brands, and media.
Even the most striking work will struggle to find an audience if its background and themes remain unknown.

Layer 5: Building a Sustainable Practice

Once you achieve a certain level of recognition, you need the ability to design a sustainable practice. The focus shifts from temporary buzz to building a foundation that supports long-term expression and growth.
This includes forming alliances, branding, establishing a workspace or team, managing finances, and preparing for potential public backlash or copyright disputes.
As your scope grows, so do the decisions required outside the art itself. You don’t have to handle everything alone; collaborating with others becomes crucial.
AI Creators aims to support not just the connection between art and opportunity, but the long-term growth and sustainability of these artists.

Why Technology Alone Does Not Make an “Artist”

Technical proficiency alone cannot guarantee artistic value. Value lies not just in the output itself, but in the questions behind it and the editorial decisions that shape it.

What Does a Generative AI Artist Create?

Business Insider Japan

The domain of a Generative AI Artist is no longer limited to a single static image. Their scope has expanded significantly into video, audio, physical spaces, IP, and product design.

Static Images, Fine Art, and Photographic Expression

High-fidelity image generation remains a vital foundation. However, these are increasingly presented not merely as illustrations, but as fine art or conceptual photography.

Video, Music Videos, Animation, and Short Films

With the advancement of video generation AI, time-based media production has expanded significantly. Attempts to reconstruct visual grammar itself—through music videos, short films, and animated expressions—are underway.

Music, Audiovisual, and Live Performance

In addition to music production using audio generation models, audiovisual expressions and live performances that synchronize video and sound are becoming critical domains.

Character IP and Worldbuilding

Consistently generating characters and fictional worlds allows for an approach focused on IP development. This area aligns closely with Japanese character and narrative culture.

Exhibitions, Installations, and Spatial Experiences

There is a growing movement to deploy generated outputs into physical spaces, evolving into immersive installations and spatial experiences. This involves complex expressions combining projections, spatial audio, and physical 3D outputs.

Integration with Fashion, Crafts, Architecture, and Products

Generative AI outputs are utilized as inspiration for pattern design and product development. Connecting AI with material domains like crafts, fashion, and architecture represents a major future opportunity.

A Roadmap for Generative AI Artists

Here, we outline the progression for beginners to deepen their practice as Generative AI Artists across five stages. While individual paths vary, this serves as a broad conceptual map.

Stage 1 (0–3 Months): Exploration and Immersion

Begin by focusing intensely on one or two tools. Through volume and repetition, observe what naturally draws your interest. The goal is to be able to articulate the “reason I create this” in a single sentence.

Stage 2 (3–9 Months): Style and Experimentation

Next, cultivate the sense of building an ongoing body of work rather than generating one-off pieces. Establish your creative axis by focusing on motifs, colors, and compositions. When necessary, experiment with custom models or LoRAs to learn the balance between reproducibility and variation.

Stage 3 (9 Months–2 Years): Publishing and Contextualization

At this stage, begin presenting your work publicly. Managing social media, building a portfolio, submitting to open calls and exhibitions, and drafting artist statements (often in English) become essential. You are now entering a phase where you must consider “how the work is read.”

Stage 4 (2–4 Years): Specialization and Crossover

Once a definitive style and theme emerge, expand into other domains. Explore where your expression connects—be it exhibitions, video, spatial design, corporate collaborations, education, or research. The key is not to be someone who “can do anything,” but to build a reputation where people say, “this person is the definitive voice in this specific area.”

Stage 5 (4 Years+): Definition and Legacy

Ultimately, you may progress to a position where you help define the field itself, rather than just publishing your own work. This stage involves shaping the wider ecosystem through critique, education, community building, and mentoring the next generation.

Technical Mastery is Not the Only Goal of the Roadmap

Tools will continue to evolve. What remains is what you observed and the body of work you accumulated. The essence of this roadmap lies in sustaining your core inquiries, rather than merely chasing the latest technical updates.

Complex Monetization Models for Generative AI Artists

The activities of a Generative AI Artist do not rely solely on direct artwork sales. It is vital to construct a multi-layered practice combining one-off revenue, recurring revenue, and trust building.

Artwork Sales (Physical, Digital, Editions)

Selling physical prints, digital works, or limited editions. This is a foundational revenue stream for building an artist’s reputation.

Commissioned Work (Advertising, Music Videos, Visual Production)

Producing key visuals, videos, and music videos upon request from corporations, brands, and other artists. These are sometimes structured as true collaborations rather than standard outsourcing.

Brand Collaborations and Corporate Tie-Ups

Connecting the artist’s style and themes with a brand’s worldview or campaign. This generates not only one-off compensation but also long-term visibility and credibility.

IP, Character, and Licensing Operations

Expanding self-developed characters and worldviews into intellectual property for licensing, publishing, and merchandise. This holds the potential for a sustainable recurring revenue base.

Education, Speaking, Workshops, and Grants

Alongside production, activities can include lectures, workshops, academic speaking engagements, and securing cultural grants. This builds both revenue and institutional credibility.

Why “Artists Can’t Earn” is an Outdated View

Modern Generative AI Artists are well-positioned to diversify their revenue streams by combining sales, commissions, IP, corporate collaborations, and education. The key is not to rely on a single method, but to construct a business model that naturally aligns with your creative expression.

Why Corporations and Brands Seek Generative AI Artists

Zalando | In Q4 2024, roughly 70% of editorial campaign assets will be AI-generated

Why a Standard AI Operator is Insufficient

When companies seek new brand experiences and new forms of expression, merely generating outputs according to instructions is often inadequate. What they require is a creative voice who possesses a unique perspective, clear themes, and the ability to design meaning.

How Vision-Driven Expression Elevates Brand Value

Collaborating with Generative AI Artists can signal a company’s forward-thinking and cultural awareness. It serves not just as advertising, but as an opportunity to update how the brand communicates its story.

Applications in Advertising, Video, Spatial Design, and IP Development

Generative AI Artists can contribute across a wide range of scenarios, including key visual production, brand videos, event space design, and character development. They are particularly effective when a brand requires a level of originality that is difficult to achieve through traditional production methods alone.

5 Points Companies Should Verify Before Commissioning

Does the artwork’s worldview connect with the company’s brand?
Is the transparency of the generation process ensured?
Is there a clear stance on copyright and training data?
Is the expression sustainable, rather than relying on temporary buzz?
Can the artist manage practical communication and delivery schedules?

Current Landscape: Global vs. Japan

AI Art Markets, Exhibitions, and Criticism Abroad

Internationally, the movement to treat generative AI expression within the contexts of auctions, exhibitions, and media art is expanding. The discussion is advancing beyond market value to include institutional critique and the artwork’s theoretical grounding.

Expanding Creative Possibilities in Japan

Japan offers a highly compatible foundation for generative AI expression, driven by strong traditions in animation, character culture, video editing sensibilities, and deep narrative structures.
Japanese creative output naturally tends toward uniqueness in its handling of narratives, characters, textures, and semiotics.

Why Generative AI Expression from Japan Tends to Be Unique

Japanese creative output naturally tends toward uniqueness in its handling of narratives, characters, textures, and semiotics. This is not a matter of direct superiority over international markets, but rather an environment that inherently fosters differentiation.

Navigating Copyright and Ethics

Fundamental Issues of Copyright and Training Data

Copyright for generative AI works remains a subject of ongoing debate globally. Reports published in January 2025 by the United States Copyright Office (USCO) emphasize the importance of human creative contribution.
(Source: U.S. Copyright Office)

In Japan, the Agency for Cultural Affairs has also outlined key discussion points, focusing heavily on the presence of human creative intent and expression.
(Source: Agency for Cultural Affairs, Japan | Regarding AI and Copyright)

Issues of Imitation, Style, and Transparency

Excessively imitating the style of a specific artist should be approached with caution, not only from a legal standpoint but also considering authorship and ethics. The ability to explain the data and methodologies used in production is directly tied to an artist’s credibility.

How to Protect and Expand Your Authorship

Cultivating custom models, proprietary data, and unique production workflows helps protect your creative output. For instance, Adobe’s Firefly Custom Models provides an environment for eligible plans to consistently train and generate specific, proprietary styles.
These practices serve not merely as defensive measures, but as expansions of authorship.

Considerations for Corporate Projects

In corporate commissions, it is necessary to establish clear terms regarding copyright ownership, usage rights, modification parameters, and accountability for the generation process in advance.
Designing the contract is just as important as the production itself.

Authorship in the Era of AI Co-Creation

Moving forward, the focus will likely shift from whether AI was used, to the types of human judgments made and the location of the creative contribution. Authorship emerges not just from the content of the final piece, but from the accumulation of the production process and editorial decisions.

Future Prospects for Generative AI Artists

The Commoditization of “Operators” vs. The Rising Value of “Artists”

As generative AI tools continue to improve, standard outputs will become increasingly difficult to differentiate.
Conversely, artists with clear themes and a recognisable body of work are likely to be valued much more highly.

What Remains After the Tools Become Transparent

Eventually, the use of generative AI will no longer be considered special. When that time comes, what remains will be what the artist observed, what they selected, and the context in which the work was presented.

The Ideal Creator in the Era of AI Agents

In the future, the weight of editing, staging, and creative direction is expected to increase further. As AI takes on multiple stages of production, humans will be required to possess the judgment to decide “what must be established.”

Shared Traits of Next-Generation Cultural Leaders

Individuals who possess their own inquiries and can translate them into a sustained body of work—beyond mere adaptability to new tech—are the ones expected to lead the creative field moving forward.

Frequently Asked Questions (FAQ)

Q. Should I aim to be an AI Creator or a Generative AI Artist?

If resolving business challenges and building a practical career is your priority, aiming to be an AI Creator is the realistic path.
However, if you want to center your practice around your own themes and strong expressive impulses, the path of a Generative AI Artist may be more suitable. The two are closely related, and dual roles or transitions are highly plausible.

Q. Can I become one even if I cannot draw or program?

Yes, it is possible. However, skills such as verbalizing prompts, discerning generated results, and designing the conceptual framework of the artwork are absolutely essential.

Q. Where should I start?

Begin by selecting a tool that suits you and interacting with it daily. Simultaneously, observe pioneering examples globally and try to articulate in words why those specific expressions are successful.

Q. If my work is strong, how do I connect it to exhibitions and commercial jobs?

Beyond the quality of the artwork, you need a portfolio, clear statements, consistent visibility, and networking.
If it is difficult to manage everything alone, utilizing communities or platforms that help connect artists with exhibition and collaboration opportunities is highly effective. Peer reviews and industry conversations serve as crucial catalysts.

Q. Will the copyright of my artwork be recognized?

This cannot be answered universally.
Generally, works generated solely by AI without human input are often denied copyright. Conversely, workflows that demonstrate clear human creative contribution—such as underlying philosophy, prompt design, structuring, selection, retouching, use of proprietary data, and research—are much more likely to be treated as copyrighted works.
Ultimately, decisions vary based on national regulations and specific cases.
Refer to AI Governance – 3. Human-Centered Creative Principle for more details.

Q. What is required to work internationally?

The ability to articulate your work in English is essential. Submissions for international art festivals and contests are predominantly in English.
Beyond the visual impact of your art, a statement conveying your themes and intent serves as the common language connecting you with overseas galleries and collectors.

Conclusion

A Generative AI Artist is not simply someone using AI to make images.
They are creative voices offering unique inquiries and sensibilities to society through the new medium of AI.

As generative AI capabilities continue to improve, differences in the tools themselves will become imperceptible.
Ultimately, what remains is not which model was used, but what was observed, what was selected, and the context in which it was materialized.

Therefore, the path of a Generative AI Artist is not merely a race to keep up with the latest tech. It requires refining technical skills while simultaneously deepening personal inquiries, cultivating a worldview, and developing a language that connects with others.
Over time, this ongoing commitment is what turns someone from a skilled operator into a true artist.

Even as tools change, the expression itself remains. And the questions and intent residing within that expression will ultimately determine your value in the era to come.

Taking the Next Step

To sustain a creative practice, you need not only to refine your work but also to build pathways that deliver it to the right audience.
How you structure your environment for learning, publishing, and securing collaborations is an integral part of your artistic journey.

en.ai-creators.tech

Generative AI Courses & AI Creator Network

https://en.ai-creators.tech/personal/

Learn practical generative AI skills, build your portfolio, or apply to join our global AI creator network and connect with new creative opportunities.

en.ai-creators.tech

Generative AI Creative Services for Business

https://en.ai-creators.tech/enterprise/

AI Creators supports companies with generative AI creative production, AI creator matching, corporate training, strategic advisory, PR, and partnerships.

The post What Is a Generative AI Artist? Skills, Monetization & Roadmap first appeared on AI Creators.

Midjourney V8 Alpha: What Native 2K Means for Creative Workflows

aratama 璞 | Editor-in-Chief, AI Creators — Thu, 19 Mar 2026 19:35:40 +0000

Update Overview: Released exclusively on Midjourney’s web alpha on March 17, 2026, V8 Alpha delivers generation speeds roughly 4 to 5 times faster than previous versions, along with improved text rendering.
Key New Feature: The new “–hd” parameter allows for the direct generation of native 2K (2048px) images, bypassing the need for a separate upscaling process.
Core Evolution & Considerations: Beyond raw image quality, the model’s prompt adherence (steerability) has significantly increased. However, features like “–hd” consume four times the standard GPU time, requiring strict yet practical cost management.

Signs of a Redefined Standard in Image Generation

Until now, the standard workflow for generative AI imaging has been to generate multiple low-resolution concepts and upscale the best ones later. However, it may be time to rethink that approach.

Midjourney’s next-generation “V8 Alpha” model, released in preview on March 17, 2026, suggests a meaningful shift in how creators balance speed, resolution, and computational cost. In this article, we will organize the facts surrounding the latest features based on official announcements and unpack the structural impact on practical workflows through real-world impressions from the field.

News Overview: What Changed in Midjourney V8 Alpha?

Today we're starting to test an early version of our V8 model with our community. It's much better at following prompts, 5x faster, has native 2K modes, improved text rendering and the best personalization, sref, and moodboard performance ever. Have fun! pic.twitter.com/bc54Iod3nv

— Midjourney (@midjourney) March 17, 2026

According to Midjourney’s official announcements and documentation, V8 Alpha is currently available exclusively on “alpha.midjourney.com” and is not yet supported on Discord or the main site. The primary changes in this update are as follows:

Dramatic Speed Increase: The generation speed for standard jobs is now roughly 4 to 5 times faster compared to previous versions.
Native 2K Generation (–hd): The newly introduced “–hd” parameter enables the direct generation of 2048-pixel images without going through an upscaler.
Improved Text Rendering: By using quotation marks in your prompt, text elements within the image are now rendered with greater accuracy.
Costs and Limitations: Relax mode is not currently available in the alpha (Fast mode only). Additionally, “–hd” and the high-quality “–q 4” parameter each consume 4 times the standard GPU time. For now, combining these parameters with style references (–sref) or the moodboard feature is disabled.

Voices from the Field: Evolving from “Vibe-Driven” to a “Controllable Pro Tool”

While the improvements on paper are impressive, what matters most in actual production is how the nature of the tool has changed. Through initial testing and community verification, the intrinsic value of V8 Alpha is gradually becoming clear.

Tried Midjourney V8α a bit.

My first impression is that it feels much more controllable.
Prompts seem to translate into images more directly than before, which makes the model easier to steer.
That said, I do not feel the images themselves suddenly became dramatically better.
It… pic.twitter.com/u7i9ECoQSe

— aratama 璞 (@aratamadao) March 18, 2026

Practical Implications

From Magically Beautiful to Accurately Responsive: Previous versions of Midjourney tended to generate aesthetically pleasing images based more on broad aesthetic cues, even with short prompts. However, V8 Alpha reacts much more faithfully and directly to specific instructions.
A Phase That Tests Creators’ Articulation Skills: Rather than a massive leap in raw image quality, the tool’s steerability has clearly improved. While this makes it easier to position elements and compose shots for commercial work, it also suggests that output quality will depend more heavily on the creator’s ability to give clear, precise direction.

Structural Shifts: Blurring the Line Between Drafts and Final Assets

By combining faster generation speeds with native 2K output and high prompt steerability, the prevailing multi-step process—brute-forcing iterations at low resolutions and upscaling the successful compositions—could become less necessary in some use cases.

Being able to iterate exactly as intended at a resolution close to the final output naturally narrows the gap between draft-stage ideation and near-final output. This signals a transition to a new production foundation that allows for near real-time iterations.

AI Creators Score (Proprietary Evaluation)

The AI Creators editorial team evaluated the impact of this update on the creative industry across four axes.

Overall Score: 90 / 100
Impact: 9 / 10
The substantial increase in iteration speed and prompt adherence (steerability) has the potential to significantly shorten production timelines in meaningful ways.
Novelty: 8 / 10
While direct 2K generation is highly valuable for practical work, it is more evolutionary than revolutionary—a natural and powerful progression of the diffusion model architecture.
Practicality: 9 / 10
The ability to easily craft intended compositions offers immediate benefits to all commercial users. On the flip side, it introduces new operational challenges regarding GPU cost management associated with higher resolutions.
Momentum: 10 / 10
Active verification by the community began immediately upon release, and discussions among practitioners regarding the fundamental shift toward better steerability remain highly active.

Decision Memo: Key Considerations and Risks

For enterprises and teams using Midjourney, we have compiled the recommended actions to take in the near term.

Immediate Action (Audit): Review GPU consumption patterns in Fast mode across your teams. If necessary, establish internal usage rules for high-cost parameters like “–hd” (which consumes 4x the usual resources).
Monitor: Keep a close eye on the implementation timeline for Relax mode and the point at which “–hd” becomes compatible with existing features like style references, leading up to the official V8 release.
Risk Management (Manual Update): Assess the risk of rapidly depleting GPU hours through the unintentional overuse of high-cost parameters, and update internal prompt operation and production manuals accordingly.

AI Creators Insight

With the integration of the generation process and high-fidelity output, combined with improved AI responsiveness, the workflow may be shifting from vibe-led prompting to more direct, high-fidelity creative control.

The practical benefits of this technological evolution extend beyond mere speed. Honing your own directional precision while optimally balancing cost and quality—this kind of operational discipline is likely to become increasingly important as a core skill for AI creators moving forward.

(*AI Creators can assist with formulating AI adoption guidelines and reviewing operational designs at the team level. Please feel free to reach out to us.)

The post Midjourney V8 Alpha: What Native 2K Means for Creative Workflows first appeared on AI Creators.