AND THE OSCAR GOES TO...ARTIFICIAL INTELLIGENCE
by Alona Elkayam
·
It feels a little too quiet in this vertiginous moment in the history of creative work and artistry. Generative AI tools have arrived. Sophisticated, affordable, astonishing and the question that copyright law is now forcing everyone to confront is not whether AI can make something that looks, sounds, and feels like art. Clearly, it can. The question is who the author in AI generated work is and whether that answer matters.
Coca-Cola, for its 2025 holiday campaign, worked with five AI specialists to sort through more than 70,000 generated clips over thirty days to remake a Christmas ad it first made in 1995. The Verge ran the headline ‘Coca-Cola’s New AI Holiday Ad Is a Sloppy Eyesore.’ System1 Group, an advertising research firm, gave it a 5.9-star rating out of 6 for emotional impact, the highest score its model awards and the company’s own data showed it outperformed previous campaigns with certain audiences. Poor use of AI. Nike, working with AKQA studios across Melbourne, Portland, and São Paulo, used machine learning to pit a 17-year-old Serena Williams against a 35-year-old Serena Williams in a virtual match simulating 130,000 games from archival footage across two decades of her career. Smart use of AI. The live final drew 1.7 million YouTube viewers, broke Nike’s organic-view records, and won the Digital Craft Grand Prix at the 2023 Cannes Lions Festival of Creativity. Heinz asked DALL-E 2 to visualize ‘ketchup’ and found that the AI conjured its bottle so reliably even when prompted with ‘Renaissance Ketchup Bottle’ and ‘Ketchup Tarot Card’ the campaign became a viral argument for the brand’s cultural dominance. Meh use of AI.
The gap between what these tools generate and what the human eye or ear can distinguish from reality is narrowing faster than any regulatory body has managed to track. At a talk about AI on March 12, 2026 in New York at Columbia Business School, venture capitalist Erica Duignan Minnihan of Reign Ventures reminded the audience that government regulation is typically ten years behind. While California signed the AI Transparency Act in September 2024 requiring major AI providers to embed provenance disclosures in generated content, and New York signed a law in December 2025 requiring advertisers to disclose when AI-generated synthetic performers appear in commercials, a White House Executive Order issued that same December immediately sought to preempt all state-level AI regulation in favor of a federal standard that does not yet exist. House Republicans have proposed a ten-year moratorium on state AI regulation. America’s lack of regulation is driven by a capitalistic race with China that has our government as amped to win as Colin Jost’s caricature of Pete Hegseth. In Europe, the governing bodies of AI technology is more maternal. The European Union will require watermarking of AI-generated content by August 2026. The United States has no equivalent law on the horizon. European law also has a blanket law against deep fakes and other enshittification tools that America is, for now, not bothering with. And it’s causing issues.
The Volkswagen campaign that resurrected Elis Regina, the Under Armour film that most viewers never identified as AI-made, or the deepfakes of public figures circulating across social platforms were not required to announce themselves. Generative AI does not come with a confession. Which means the question of who made something is no longer merely legal or philosophical. It is, increasingly, a question of what is real and who gets to decide.
The answer, as of early 2026, is complicated in ways that should alarm every creator, brand, and executive who has been cheerfully deploying these tools without reading the fine print.
In January 2025, the U.S. Copyright Office issued Part 2 of its landmark report on artificial intelligence and copyright. It is the foundational policy statement on which every subsequent legal development has been built. Its core finding was that work generated solely by AI, without meaningful human creative control, cannot be copyrighted. It enters the public domain the moment it’s rendered. Anyone can use it. Anyone can copy it. You own nothing.
The report did acknowledge a path forward: Hybrid works — those where a human has substantially shaped, arranged, edited, or transformed AI output may qualify for copyright protection. But only for the human-authored portions. The AI’s contribution, even when it makes up ninety percent of the finished product, remains in the legal ether, unclaimed.
In March 2025, the U.S. Court of Appeals for the D.C. Circuit affirmed the lower court’s ruling in Thaler v. Vidal. At the heart of this case was the question if a machine be an author. The answer, again, was no. The case went to the Supreme Court. And on March 2, 2026 as this essay was in its final edits, the Supreme Court declined to hear the appeal. It is the most current data point in this entire piece, and arguably the most important: the highest court in the United States has now effectively confirmed that the door on AI-only copyright claims is closed.
The man at the center of the case, computer scientist Stephen Thaler, created an image in 2012 using an AI tool he had himself built, called DABUS, and titled it ‘A Recent Entrance to Paradise.’ He applied for copyright in 2018. The Copyright Office rejected it on the grounds that copyright requires human authorship. A district court upheld the decision. The D.C. Circuit upheld it again. And now the Supreme Court has declined to intervene refusing even to hear the argument.
The implications are not abstract. They run directly into the workflows of advertising agencies, film productions, music studios, game developers, and anyone else who has begun treating these tools as a creative shortcut rather than a creative collaborator. IP attorneys at Skadden, Arps noted in a May 2025 analysis that even AI training itself carries risk and using copyrighted works to build these models may constitute infringement, opening the door to liability that flows in multiple directions at once.
The American Bar Association, analyzing two significant June 2025 rulings. Kadrey v. Meta Platforms and Bartz v. Anthropic summarized the new terrain bleakly for brands. Businesses using AI-generated content bear the same copyright liability as if a human had authored the material, but without the protections human authorship would normally confer. An advertising agency using AI to create campaign visuals that resemble existing work faces exposure even if the AI generated the resemblance without being asked to, and even if no one at the agency knew it happened.
Michael Martensen, the founder of Martensen IP Law and one of the most widely-cited voices on AI copyright risk, frames the practical standard that should be posted in every creative department in America: treat AI output the way you would any third-party content. Check it, edit it, document your role in shaping it, and be prepared to prove your creative contribution in court. Paying for an AI tool does not confer copyright. It confers access. Under current law, any AI-assisted work seeking copyright protection must clear three hurdles.
-
The human creative contribution must be substantive, meaning purposeful direction, editorial selection, and original arrangement, not a prompt alone
-
Documented with timestamped records of every revision and decision
-
Provable because in any copyright dispute the burden of demonstrating human authorship falls entirely on the creator, not the machine.
To understand the gap between what these tools offer and what they guarantee, it helps to read the terms of service which most users do not. The major platforms vary in what they promise. Suno, the AI music generator, grants commercial ownership of output on paid tiers but stops well short of guaranteeing copyright in the legal sense. Runway Gen-4, the video generation and editing tool, offers generous commercial rights including use at film festivals and on YouTube, but cannot guarantee that unedited AI clips meet the human-authorship threshold for copyright protection. Midjourney permits commercial use on paid plans, though Disney and Universal have pending lawsuits over its training data, and it provides no indemnification for infringement. Google’s Veo 3 allows commercial use on its AI Ultra tier as part of the broader Flow filmmaking suite. ElevenLabs, the voice synthesis platform, raises specific exposure under the ELVIS Act, signed into Tennessee law in 2024, which extends protections to digitally replicated voice performances. Adobe Firefly stands apart from the rest: trained exclusively on licensed content and public domain works, it is the tool most IP attorneys now point to when brands ask which platform carries the least legal risk. The tradeoff is stylistic constraint — Firefly’s outputs tend toward the commercial and the competent rather than the genuinely surprising. For companies prioritizing IP cleanliness over aesthetic ambition, it is increasingly the default choice.
The distinction between ownership and copyright is the one that catches most users off guard. Suno’s documentation makes it clear: users own the output commercially. They can sell it, license it, post it. What they cannot necessarily do is stop someone else from using it. Copyright is the right to exclude. Ownership, as these platforms define it, is merely the right to use. For a brand building campaign assets or an artist building a catalog, that gap is the difference between owning and renting.
Not everyone is outsourcing the future of humanity to the frontier labs. Guillermo del Toro, the three-time Oscar-winning director of Pan’s Labyrinth and The Shape of Water, has been perhaps the most eloquent voice of opposition from within the creative establishment. Promoting his Netflix adaptation of Frankenstein in October 2025, he was asked by NPR about his stance on generative AI.
I am not interested, nor will I ever be interested. I’m 61, and I hope to be able to remain uninterested in using it at all until I croak. ‘I’d rather die.’
— Guillermo del Toro, NPR, October 2025
Del Toro’s objection is not merely aesthetic, though it is that too. At the British Film Institute (BFI) the previous year, he had dismissed AI image generators as capable of producing ‘semi-compelling screensavers.’ His deeper argument is philosophical: that the value of art is inseparable from the risk involved in making it. ‘The value of art is not how much it costs and how little effort it requires,’ he said in London. ‘It’s how much would you risk to be in its presence?’
For his Frankenstein, del Toro requested ‘real sets, old-fashioned craftsmanship with people painting, building, hammering, plastering.’ He sees in the tech industry’s rush toward AI the same blind hubris he sees in Mary Shelley’s Victor Frankenstein: a willingness to create without adequately considering what has been created, or why, or at what cost. ‘My concern is not artificial intelligence,’ he told NPR, ‘but natural stupidity.’
He is not alone. Steven Spielberg told Reuters in June 2025 that he does not want AI making any creative decisions he cannot make himself. DC Comics president Jim Lee has stated the company will not support AI-generated storytelling or artwork. These voices carry enormous cultural weight about what creativity is, and whether a tool that requires no risk, no failure, no personal investment can produce work that matters.
Jon Lam is a senior storyboard artist at Riot Games, whose credits include Diablo 4, Overwatch 2, Star Trek: Prodigy, and Invincible, and who founded a movement called CreateDontScrape.com. His Instagram bio reads simply: ‘🚫AI.’ He is not against technology. He is against what he describes as the harvesting of working artists’ livelihoods to train systems that will then replace them. ‘How many times have we told generative AI companies and their users to stop hurting us,’ he wrote on LinkedIn, ‘only to be met with: We’re not hurting you. Adapt or die. It’s an amazing tool. This tool is wearing our faces.’ What Lam and artists like him are naming is not technophobia. It is an economic grievance. The training data powering these systems was taken from human artists without consent, and the profits from that action flow entirely to the platforms, not to the people whose work made the models possible.
Justine Bateman, filmmaker, computer science graduate, and SAG-AFTRA’s advisor on AI during the 2023 strike, has built perhaps the most sustained institutional response. She founded the CREDO 23 Film Festival, a no-AI-allowed event launched in Los Angeles in 2025 that she describes as ‘a tunnel for human artists through the theft-based, job-replacing AI destruction.’ Speaking to the BBC in September 2025, she inverted the industry’s favorite warning about falling behind: ‘You will absolutely be left behind if you’re using AI. You completely stop your forward momentum as an artist.’ Her argument is not sentimental. It is strategic. Generative AI, she has said repeatedly, ‘will not automate anything exceptional.’ For a creative economy that has been suffering from a volume-over-originality problem for fifteen years (Star Wars sequels anyone?), she argues, AI is accelerant, not cure.
The question of who the tools empower and who they displace does not have a tidy answer. But it is worth sitting with, because the human artists most harmed by unchecked AI deployment are often the same people whose labor, judgment, and risk make any finished work worth owning or knowing in the first place.
In November 2025, Imagine announced a creative partnership with Obsidian Studio, an AI-native production company that had spent the year producing high-budget commercials for Louis Vuitton, Longchamp, Crayola, the NBA, and ESPN. The partnership would allow Obsidian to expand into narrative feature film — with Howard’s creative involvement and institutional backing. It is, in its way, a model for the hybrid approach: human artistic vision directing AI-powered execution.
For the world of marketing and advertising, the practical implications are significant. An advertising agency producing AI-generated content that resembles existing campaigns faces potential liability even if the outputs serve different purposes. Disney’s complaint about Microsoft’s Bing AI imaging tool generating Pixar-style imagery offered an early illustration of how brand identity and AI generation can collide. The issue was resolved by tweaking the model, but these systems were trained on the visual and sonic culture of the world, and that training shows up in what they produce.
IP attorneys now recommend companies create inventories of their AI tool usage, establishing when humans must review outputs, and building documentation protocols to prove human authorship in the event of a challenge. The advice from Cummings & Cummings Law is blunt and practical: ‘Avoid simplistic assumptions such as I paid for the tool, therefore I own the content. Payment does not guarantee copyright protection or exclusivity.’
For brands that want commercial rights they can actually enforce, the calculus has become clear: human creative contribution needs to be substantive, documented, and provable. That means not just a prompt, but editing, arranging, transforming, directing. It means treating AI output as raw material rather than finished product.
Netflix has published internal guidelines that require transparency when AI generates content and prohibit its use to replace union-covered work without consent. Pixar and Toei Animation have incorporated AI tools into background generation and storyboarding. LTX Studio offers a full-stack AI filmmaking suite — storyboarding, shot planning, and video generation that functions most naturally as a pre-production tool, a way of getting ideas out of one’s head and into a visual form before the camera turns on.
At our firm, Far From Timid, we call it The Storyboard Principle: use AI where it accelerates human decision-making and reduces the cost of exploration, but bring human craft and judgment to bear on anything that will be published, distributed, or monetized. Rough out the visual world of your film in Midjourney. Prototype the music in Suno. Build your shot list with Runway. Then hire the cinematographer, the composer, the editor — and make something that is legally, aesthetically, and morally yours.
The Storyboard Principle: use AI where it accelerates human decision-making and reduces the cost of exploration — but bring human craft and judgment to anything that will be published, distributed, or monetized. AI is the greatest pre-production tool ever made. Treat it that way.
The framework, proposed here as a practical guide for brands and creators navigating this transition, is a brief for strategic clarity. AI is a pre-production tool of formidable power. It should be used as such. The storyboard has always been the place where ideas get tested, iterated, and clarified before the real work begins. Generative AI is the greatest storyboarding tool ever made. Treating it as a finished-goods factory is both a legal mistake and a creative one.
The Storyboard Principle is also, not coincidentally, what the Copyright Office seems to be gesturing toward: a world in which AI functions as an extraordinarily powerful creative instrument, the way a camera or a piano is an instrument, and in which the copyright belongs to the human player, not the machine.
Here is the uncomfortable truth at the center of this conversation. The Copyright Office is right, not merely as a matter of legal interpretation, but as a matter of culture and value. Work that requires no human risk, no human failure, no human point of view tends to look like work that required none of those things. It can be beautiful, technically proficient, and approximate emotion with striking accuracy, but it cannot surprise you in the way that human work surprises you. With the evidence of a consciousness behind it, making choices, taking chances, and getting things wrong.
Jerry Saltz, the Pulitzer Prize-winning art critic for New York Magazine and one of the most polarizing voices in the contemporary art world, made this argument unforgettably when he reviewed Refik Anadol’s AI-generated installation Unsupervisedat the Museum of Modern Art in 2022. Anadol had fed AI metadata from more than 138,000 works in MoMA’s collection and let the system reinterpret art history as a continuous flow of morphing abstractions. Visitors spent an average of 38 minutes with the piece, far longer than the 28 seconds typical for museum art. None of that impressed Saltz. He called it a ‘massive techno lava lamp’ and a ‘half-million-dollar screensaver.’ Anadol responding that ChatGPT writes better than Saltz. But Saltz’s deeper point made on 60 Minutes, where he described most current AI art as ‘an average of averages’, has proven harder to dismiss: popularity, he argued, is not a measure of artistic success. How long you spend with a work is not the same as what it asks of you.
And something remarkable is happening in response. The most sophisticated brands are doing the opposite of Coca-Cola and Nike. They are leaning into craft, materiality, and human originality not despite the AI moment, but because of it. They are treating analog as a competitive signal.
Consider what Apple did in November 2025 when it rebranded from Apple TV+ to Apple TV. It needed a new logo animation, an ident, that would appear before every film and series it produces. A studio of Apple’s resources could have generated something photorealistic in an afternoon using any number of tools. Instead, it spent weeks in a London studio with its agency TBWA\Media Arts Lab, building a physical glass version of the Apple TV logo, then filming it with macro lenses and color gels against changing light. Every shimmer in the final sequence is real — not rendered, not composited, not generated. The agency’s own description of the work: ‘craft should be felt, not faked.’ The sonic identity was composed by FINNEAS, the Oscar-winning musician. The typeface was custom-built by Apple’s own design team. The contrast with Coca-Cola’s AI-generated holiday debacle was not lost on anyone.
Every shimmer was made for real, no CG shortcuts — a nod to Apple’s belief that craft should be felt, not faked.
— TBWA\Media Arts Lab, on the Apple TV rebrand, November 2025
OpenAI, one of the companies most responsible for the current generative moment shot its first major ChatGPT advertising campaign on 35mm film. Anthropic, the AI safety company, ran a brand campaign filmed entirely with traditional analog cinematography, scored with a Madvillain soundtrack and produced with the care of a short film. These are companies that could have used their own products to make their ads. They chose not to. The implicit message was clear: human craft has become a premium signal.
And it’s not surprising these brands are using analog methods. Gen Z, the first generation to grow up entirely inside the streaming economy kicked off this trend. They’re buying vinyl records at a rate that has surprised the music industry. According to the Recording Industry Association of America’s 2024 year-end report, vinyl shipments in the United States reached 44 million units, the format’s eighteenth consecutive year of growth generating $1.4 billion in revenue and outselling CDs for the third straight year, their highest revenue level since 1984. Gen Z now accounts for 27 percent of all vinyl purchases. And 56 percent of them say they buy records for the aesthetic, not exclusively for the sound. Around 40 percent of vinyl buyers don’t even own a turntable. They are buying the object. They are buying the evidence of making.
Live Nation, for its part, reported its biggest year of concert attendance in history in 2025 surpassing even the post-pandemic 2023 boom. When everything is infinitely reproducible and available on demand, the irreducible live experience — the one that cannot be generated, streamed, or copied commands a premium. Fortune, surveying the trend in February 2026, described Gen Z’s turn toward analog as ‘an act of identity assertion and a cry for respite from the enormous noise perpetuated by the 24/7 social media cycle.’ For these consumers, the handmade is not nostalgic. It is defiant.
Brand loyalty, fandom, cultural movements do not cohere around content alone. They cohere around authorship. Around the sense that somewhere at the center of this film, campaign, record, logo, there is a person whose vision you are choosing to trust. You cannot belong to something that has no author. And in an era when the volume of generated, authorless content is growing exponentially, the work that carries a human signature is not just legally stronger, it is the only work capable of building the thing every brand ultimately wants: a world people choose to live inside.
It is not that AI-generated content cannot be effective. The Nike Serena campaign proves it can. It is that AI-generated content cannot be yours legally, aesthetically, or in the eyes of an audience that is increasingly sophisticated about the difference.
Creativity is not the production of output. It is the expression of a consciousness choosing, over and over, what matters and being willing to be wrong, and trying again. That has always been what makes a piece of work worth the audience’s time. Art softens the blow of human suffering and we must be willing to keep the standards high for what is creativity.
The copyright is the legal, moral, and artistic claim belonging to whoever is willing to do that.
(Thanks for staying this long).