OpenAI, Studio Ghibli: Of Course It's Infringement! "Style" Is Protectable
"In The Style Of" Prompts Must Trigger Artist Payments, At A Minimum
By Peter Csathy, Creative Media
Welcome to this week’s “brAIn” dump, featuring my “mAIn event” — my very deep dive analysis of last week’s epic OpenAI/Studio Ghibli controversy (some would say, AI anime art heist). Next, my “AI video of the week” — my discussion of the limits of AI “fair use” (and how the rest of the world sees it). Finally, the “AI Litigation Tracker” — updates on the key generative AI infringement cases by McKool Smith. Several major updates worth checking out (check out the full “Tracker” here).
But First …
Dear readers, if you enjoy my “brAIn,” find value in my deep-dive analyses only available here, and want to support my work — consider upgrading to my new official “brAIn-iac” club, which gets you unlimited access to “the brAIn” Archive, access to new exclusive live AMA’s, discounts to my new “AI Insider Sprint Sessions,” and the sheer joy of giving my “brAIn” a dopamine hit! I’ll send you a personal note of thanks. Or really set your neurons firing by becoming a “brAIn-iac Benefactor.” You’ll also get a private 1-1 consultation. (Discounted corporate/group rates available too.)
I. The mAIn Event - OpenAI & Studio Ghibli: Yes, Legendary Anime Artist Miyazaki’s “Style” Can Be Infringed (& Yes, He Should Be Paid)
By now, you’ve likely heard about OpenAI’s new image generator. From all accounts, it’s impressive. Greg Harrison, Chief Creative Officer of leading creative agency MOCEAN (and an AI/media voice I respect), calls it a game-changer. “It’s shocking what it's capable of,” he tells me. “It’s a level of control we’ve never seen before” — so much so that OpenAI had to shut it down temporarily due to frenzied demand, even though OpenAI leverages the vast AI infrastructure and horsepower of its partner, Microsoft.
So What’s The Big Deal?
Well, how’s this for starters? That swath of new users (and existing ones too) discovered that they can now become world-class visual artists with a simple “create in the style of” prompt or two. Even anime artists, especially one anime artist that towers above all others — Hayao Miyazaki, founder of Studio Ghibli. Miyazaki is widely known for creating an entirely bespoke style of anime that he honed over the years with painstaking attention to detail. One crowd scene alone from his film The Wind Rises reportedly took over one year to create, even though it takes up only 4 seconds of screen time. That’s the kind of creative “love” that Miyazaki puts into his art.
But last week, a massive global audience took to OpenAI’s new image generator to spit out anime “in the style of Studio Ghibli” and — voila! — they too could do what Miyazaki does, but without all that fuss and muss (in mere seconds, no training needed). And so, lo’ and behold — they did. OpenAI’s CEO Sam Altman didn’t want to miss out on the fun. So, he didn’t. He “Miyazaki-ized” his profile pic, and then gleefully boasted, “We added one million users in the last hour” after the Miyazaki mania began. How great is that? His own personal “homage,” I guess, to Miyazaki and Studio Ghibli.
Only it isn’t an homage at all. And we all know it. It’s not creative inspiration. It’s mass creative infringement and market substitution. That’s as plain as day. Gone is the need, apparently, to commission Miyazaki and his studio to do his signature bespoke anime work. Now OpenAI’s image generator will do it for you.
The Creative Money Still Flows — To OpenAI
But wait. It gets even “better.” It’s not just Miyazaki, of course. Altman has made it drop dead easy for users to never need to commission visual artists again. How cool is that? That’s the beautiful thing about tech-fueled creative progress. It’s creative democratization in action, without the pesky problem of needing to pay anyone for the creative works you need. “We’re the heroes here,” Altman would say. “It’s the Artists who want to hold you back.”
Well, actually, not quite. You still need to pay something to someone. But that “someone” is Altman and OpenAI here. That’s where your money goes. In fact, ChatGPT “in app” subscription revenue and app downloads reached an all time high during the initial Studio Ghibli stampede. So there’s that.
And then there’s this. OpenAI just announced a fresh round of $40 billion at a $300 billion valuation to fuel Altman’s AI utopia. So the money is certainly there to bring creators into AI’s big tent and share at least some of that wealth via content licensing. It’s a conscious decision not to.
Payments To Artists At Least, Right?
The tech is also there to track “in the style of” prompts and either stop them from generating outputs tied to specific Artists (like Miyazaki), or at least trigger mandatory payments to the Artist. When I prompt ChatGPT to generate an image of “Sam Altman taking an artist’s money,” for example (which I did), ChatGPT refuses. “I can’t create that image,” it tells me. So OpenAI definitely has the power to stop the Miyazaki madness.
At a minimum, OpenAI could pay individual Artists like Miyazaki every time users ask ChatGPT to generate images in their style. After all, “creative democratization” doesn’t mean there’s a free lunch. Otherwise we’d still be living in the Napster piracy days of old. The tech certainly exists to implement automatic payments to Artists. It’s just not being deployed.
Think of YouTube. It too started as a mass infringement enabling machine. But in response to industry pressure (litigation and the like), it developed Its Content ID system which: “continuously compares new [user] uploads to the references for [an Artist’s] assets. Matching videos are automatically claimed on behalf of the asset, and [the Artist’s] specified match policy is applied to the claimed videos before they are published on YouTube.” If an Artist “opts in” to allow YouTubers to use their work, then payments to that Artist can be triggered when their work is used.
Here’s another example. Compulsory licensing in music. Bands can “cover" and record another Artist’s songs without the need to directly negotiate a license. The songwriter just automatically gets paid royalties when those “covers” make money.
But “Style” Isn’t Copyrightable, Is It?
“Still, what’s the big deal?,” Altman and his Silicon Valley acolytes (including his always aiming-to-please attorneys) likely would say. “You can’t copyright an artist’s style.” In Altman’s and Silicon Valley’s view, an artist’s style — even Miyazaki’s bespoke, revered anime style — is like an “idea.” And ideas are not copyrightable as a matter of law. “They’re in the air,” in the words of one memorable Seinfeld character. Style, style, style. Yes, that’s the ticket!
This is where I press the buzzer and say, “WRONG!” That’s “style over substance” in this new generative AI context. And we all know it.
OpenAI’s Two Potential Infringements
(1) Infringement #1: Training on Miyazaki’s Works
Let’s leave the trees and step back to look at the overall forest. ChatGPT now enables all of us to be great Miyazaki anime artists, because it is virtually certain that OpenAI trained its new model on Miyazaki’s art. “We build our AI models using publicly available data,” OpenAI said in response to the Miyazaki kerfuffle. And, by now, we know what those three words — “publicly available data” — mean. Hordes of unlicensed content. Tellingly, Reuters reports that when asked about Miyazaki, OpenAI did not “respond to a request for comment on the data used to train its AI models and the legality surrounding its latest feature.” Shocking, I know.
And if that’s the case, then that’s likely copyright infringement #1 right there. The first federal court to reach that “unlicensed training” issue in Thomson Reuters v. Ross Intelligence ruled, as a matter of law, that unlicensed training is not “fair use.” (Important development: that ruling was just elevated to the 3rd Circuit Court of Appeals for review; but I’m confident the decision will be upheld for reasons I’ve laid out in several earlier newsletters.)
(2) Infringement #2: Enabling Mass Infringement of Miyazaki’s Works
“Style” (like Miyazaki’s) is absolutely copyrightable when it moves into specific artistic “expression” as reflected in the Artist’s body of work.
But Altman would surely say that OpenAI isn’t enabling its users to generate exact, infringing copies of Miyazaki’s art. Rather, its new image generator only gives users a chance to use Miyazaki’s non-copyrightable style to create their own “novel” (as in, not exact duplicates) of Miyazaki-ized images. So, no harm, no foul.
Cute argument, to be sure. But wrong again.
Literal duplication of any specific Miyazaki creative work is not necessary to find infringement. The Supreme Court has made that clear. In Harper & Row v. Nation Enterprises, for example, the Court wrote: “the fact that a work is not copied verbatim does not preclude a finding of infringement.”
And here’s the thing. Miyazaki’s bespoke anime art is almost certainly part of the content fuel OpenAI uses to power its image generator, so that users can prompt ChatGPT to spit out an infinite number of Miyazaki’s. ChatGPT becomes a Miyazaki Machine! And amongst those infinite faux Miyazakis, there’s certainly at least “some” (I’m being extremely charitable here) that cross the line into infringement due to substantial similarity to Miyazaki’s “expression” in his body of anime work.
But ChatGPT Users, Not OpenAI, Are Responsible. Right?
Wrong again!
OpenAI cannot, as a matter of law, try to cleanse itself for infringement by simply shifting the blame onto its users. There’s a broad net of so-called “secondary liability,” which can take multiple forms — contributory infringement, vicarious liability, active inducement, and willful blindness. And what OpenAI is doing here certainly sounds like one or more forms of secondary liability to me. After all, users can specifically ask ChatGPT to create images “in the style of Studio Ghibli.” And then, it does.
OpenAI doesn’t seek Miyazaki’s consent for any of this, of course, even though the anime legend has made it plain that he hates AI. “I am utterly disgusted,” Miyazaki said after being shown an early AI-generated rendition of his art. Mass “Miyazaki-ation,” in his view, cheapens his art. And he certainly has a point. Shouldn’t artists have at least some control of how their body of work is exploited?
Nor does OpenAI pay anything to Miyazaki for that fuel to power its Miyazaki Machine. It’s like driving your car to the gas station, filling up your tank, and then leaving the scene of the crime without paying a thing. I don’t think Exxon would be too happy about that.
Nor would Altman if we simply took OpenAI’s ChatGPT code to power our own alternative “ethically sourced” LLM without licensing and paying him. Funny how what works. (Hey, that’s a pretty good idea, don’t you think? A new creative AI chatbot alternative powered by the entertainment industry a la Hulu.)
OpenAI’s “Everything” Machine (& Market Replacement)
But let's step back even more and take a “meta” perspective of what OpenAI is doing here (speaking of Meta, Mark Zuckerberg has expressed outright contempt for artists). Make no bones about it. Altman wants to be everything to everybody — the one-stop source for all your generative content needs, not just anime art. But why stop there? OpenAI wants to be your one-stop source for everything — your news, your shopping, your … well, you get the idea. Altman’s ultimate end game is to efficiently “replace” all other markets to create an Uber market, if you will.
“Market replacement.” Hmm, where have I heard that before? Oh yes, the U.S. Supreme Court! The nine Supremes handed down their most recent 7-2 “fair use” pronouncement in 2023’s Andy Warhol/Prince case (about which I’ve written), in a landmark ruling that reflected rare agreement by the conservative and liberal wings. The Court in Warhol rejected “fair use” as a defense to infringement on “market replacement” grounds. Yes, even this wacky Supreme Court — this “strip away all guardrails” in the name of commerce Court — sided with the creative community (plaintiff, photographer Lynn Goldsmith) to create a new “market replacement”/market substitution rationale to reign in “fair use” gone wild. After all, everyone’s gotta make a living. Yes, even the creative community!
Warhol certainly isn’t an AI case. But perhaps even the nine Justices envisioned a time when AI would come to replace their seats on the Court, rendering their hallowed columned building obsolete (surely Elon would applaud that reality and maybe even transfer their building to DOGE). In any event, copyrightability and “fair use” are intentionally flexible concepts to address new realities. Generative AI certainly fits that bill.
The Supreme Court, in Warhol, made it clear that new contexts require new rules of the game. That’s why it essentially established a new carve out to the “fair use” defense for the sake of preserving creative economic incentives and livelihoods.
So Where’s The Industry Outrage?
Media and entertainment companies, given all of this, where’s your outrage? I still haven’t seen a single major studio or streamer push back on Altman’s latest new “Miyazaki Machine.” “Shhh. What’s that you hear?” Oh, that’s right. CRICKETS!
Some of you may think that my pushbacks in these pages are hyperbolic and motivated by a certain creative nostalgia. Or, that I tend to overstate things in the hopes of stopping the AI presses and going back to a kinder, gentler, analog time. But that ain’t what this is about at all. I’ve run several tech-forward media companies, so I fully embrace technology.
But I also call it out when technology — and the companies behind it — have gone rogue by not seeking consent from, and paying compensation to, artists and rights-holders. And make no mistake. Unbridled, unlicensed AI tech is running amok and steamrolling the creative community right now, much like unbridled new tech has in the past. YouTube’s genesis, anyone? Until called out by the entertainment industry both in the press and in the courts, YouTube also enabled mass copyright infringement with impunity. And here’s the thing. It’s all happening now much faster than you may think.
So if not now with industry outrage, when?
Where The Industry Stands Today
MOCEAN’s Greg Harrison (who writes a great “Future of Creativity” newsletter) speaks eloquently about it all. He tells me that major brands, for the most part, are absolutely fine with this new AI-supercharged creative world order. Hence, the AI-generated Coke ad that flooded our screens last December (about which I recently wrote). Major studios and streamers, not so much though. They are concerned about both creative community and legal blowback, especially since every country has its own copyright laws.
Of course they are “concerned!” And they should be. But if that’s the case, then where’s the action? Silence certainly ain’t golden, especially when that gold goes only one way — away from the creative community.
We, the media and entertainment industry, are built on the backs of creatives. And at a certain point, when things go too far and cross the line — as they do here with the theft of Miyazaki’s bespoke style to enable infinite Miyazaki’s — it is “fundamentally an affront to creators,” Harrison tells me. And he, like me, is no anti-tech activist. “I welcome adoption of AI,” he says. “But it must be through the lens of creativity.”
Harrison lays it out this way:
“The underpinnings of how AI is trained are fundamentally wrong. It’s like breaking a social contract and it’s disturbing. I’m feeling this cultural perspective. AI has digitized creative minds. Not just all the ‘data.’ The best ‘data.’ And it’s all put behind a paywall to compete against the creative community.”
So What Can We Do About It?
There’s no easy answer, of course. It’s downright daunting, as Harrison points out. “The ethical, legal and creative infrastructure is just not up to understanding the implications of generative AI.”
Well, here’s one thing the media and entertainment industry can do for starters. Inspired by what unlicensed AI does, let’s steal a page from what we’re told to do in the face of potential threats. “See something, say something.” Demand that Artists get paid!
And at least they’ve finally begun to say “something,” albeit a watered-down something. It was just revealed that the Motion Picture Association of America (MPA) sent a formal comment to the Trump administration, saying that AI “can, and must, coexist with a copyright system that incentivizes the creation of original expression and protects the rights of copyright owners.” The MPA is the trade group that represents Amazon Studios, Netflix, Paramount, Sony Pictures, Universal, Disney and Warner Bros.
Not sure exactly why the MPA appears to be almost sheepish about speaking out (in Deadline’s words, the submission was “quietly posted to the MPA’s website”) — but hey, it’s a start.
Here’s the ultimate solution. Generative AI developers, take away the friction that holds back adoption of your AI chatbots and search. Exit that endless loop of copyright litigation. Work with the creative community that makes the content you need to fuel your LLMs, not against them. Hear them out. It’s surprising how effective a little outreach and listening can do. Then craft “win win” licensing deals — including potential compulsory licensing — that work for both media and tech.
There’s certainly plenty of money to go around. OpenAI’s latest $40 billion round is proof positive of that.
[What do you think? Send your thoughts to peter@creativemedia.biz]
II. AI Video of the Week: AI & “Fair Use” in the U.S. (& Overseas)
I was recently interviewed by duo AI/media mavens Ricky Sutton from the land down under and Alan Chapell of NYC on their Future Media podcast. We discussed AI “fair use” in the U.S. in light of a federal court’s recent rejection of it in Thomson Reuters v. Ross Intelligence. We also discussed how and why the U.S. stands alone in embracing “fair use” as a formal legal defense to infringement. It’s fascinating to hear a global perspective of it all, and I think you’ll get a lot out of it. Watch/listen by clicking on the video below.
III. AI Litigation Tracker: Updates on Key Generative AI/Media Cases (by McKool Smith)
Partner Avery Williams and the team at McKool Smith (named “Plaintiff IP Firm of the Year” by The National Law Journal) lay out the facts of — and latest critical developments in — the key generative AI/media litigation cases listed below. All those detailed updates can be accessed via this link to the “AI Litigation Tracker”.
The Featured Updates:
(1) The New York Times v. Microsoft & OpenAI
(2) Kadrey v. Meta
(3) Thomson Reuters v. Ross Intelligence
(4) In re OpenAI Litigation (class action)
(5) Dow Jones, et al. v. Perplexity AI
(6) UMG Recordings v. Suno
(7) UMG Recordings v. Uncharted Labs (d/b/a Udio)
(8) Getty Images v. Stability AI and Midjourney
(9) Universal Music Group, et al. v. Anthropic
(10) Sarah Anderson v. Stability AI
(11) Raw Story Media v. OpenAI
(12) The Center for Investigative Reporting v. OpenAI
(13) Authors Guild et al. v. OpenAI
NOTE: Go to the “AI Litigation Tracker” tab at the top of “the brAIn” website for the full discussions and analyses of these and other key generative AI/media litigations. And reach out to me, Peter Csathy (peter@creativemedia.biz), if you would like to be connected to McKool Smith) to discuss these and other legal and litigation issues. I’ll make the introduction.
About Peter’s Firm Creative Media
We represent media companies for generative AI content licensing, with deep relationships and market access, insights and intelligence second to none. We also specialize in market-defining strategy, breakthrough business development and M&A, and cost-effective legal services for the worlds of media, entertainment, AI and tech. Reach out to Peter at peter@creativemedia.biz to explore working with us.
Send your feedback to me and my newsletter via peter@creativemedia.biz.