The Secret Weapon Against AI Dominance
More than 90 lawsuits have been filed by creators against AI companies for copyright infringement. Authors, musicians, visual artists, and news publishers have all accused firms such as OpenAI, Meta, and Anthropic of using their copyrighted works to train AI models without permission. (The Atlantic is involved in one such lawsuit, against the AI firm Cohere.) These cases are frequently framed as the defining fight over the future of creative labor and the entertainment industry as a whole. As one of these lawsuits put it, artists are seeking to end “infringement of their rights before their professions are eliminated by a computer program powered entirely by their hard work.”
But the future of creative labor will more likely be decided through a different question within copyright law, one that has received far less attention: To what extent should AI-generated works receive copyright protection at all? In a 2024 case, Thaler v. Perlmutter, the Court of Appeals for the District of Columbia held that a work generated autonomously by an AI system cannot be protected by copyright, because copyright requires a human “author.” The Supreme Court declined to review that decision in March. With the lower-court decision left in place, the question now becomes how much AI content can be incorporated into a work before it becomes mostly or totally uncopyrightable; courts have not yet weighed in on this but may soon.
The Thaler decision (and any future decisions that refine it) will have major economic consequences for the creative industries and the workers they employ. That’s because entertainment and media companies are in the business of monetizing intellectual property. Studios license films for streaming, theatrical distribution, merchandising, and franchising. Record labels license recordings for streaming, movie soundtracks, and sampling. Book publishers license rights across formats and languages, and for television and film adaptation. Copyright protection is the engine that makes all of this run. Without it, anyone could copy, distribute, or adapt a work for free, and the entire financial model would collapse.
This means that copyrightability is necessary for profits. And this fact has quietly created a powerful financial incentive for the entertainment industry to keep humans in the loop. Even as AI-generated content has started flooding platforms such as YouTube and TikTok, we haven’t seen it migrate to the established gatekeepers: Hollywood studios, large record labels, book publishers. Netflix’s own production guidelines warn creators not to use AI to “generate main characters, key visual elements, or fictional settings that are central to the story without written approval.” Hachette recently pulled the book Shy Girl after allegations surfaced that portions were AI-written.
[Alex Reisner: The hypocrisy at the heart of the AI industry]
These are not acts of charity toward the human authors who might complain about AI use; they’re acts of business pragmatism. Cutting out human creators in favor of AI could save producers enormous sums of money. But as long as the prohibition on copyrighting AI-generated content holds, major studios, labels, and publishers must continue employing human screenwriters, actors, illustrators, songwriters, and recording artists. Not because they want to, necessarily, but because maintaining strong copyright protection allows them to license content, compete against other industry players, and stop piracy.
It’s worth acknowledging what copyright cannot do. Some creative industries will probably not survive the arrival of generative AI regardless of whether AI-generated material is copyrightable or not. Stock photography is a clear example; if a company can generate a perfectly adequate image for its website or marketing materials, it has little reason to pay a commercial photographer or license from a company such as Getty. The argument we are making applies most forcefully to the core industries where large companies still serve as intermediaries between creators and consumers: film, television, music, and book publishing. Curation—that is, trusting an intermediary to help you determine what’s worth your time—has remained valuable even as tech companies have made it easy for anyone to self-publish an album or a novel. The growing flood of AI-generated slop seems to have only intensified that demand for curation.
The recent sudden and stunning collapse of OpenAI’s video-generation tool, Sora, is a case study in why such content companies likely won’t be abandoning human authorship to embrace AI wholesale anytime soon. At the end of last year, OpenAI announced a “landmark” licensing agreement with Disney, which would give users the ability to “bring beloved characters from across Disney’s brands to Sora.” Just a few months later, OpenAI announced that it was pulling the plug on Sora altogether.
Some writers and experts suggested that Sora’s demise stemmed from its lack of popularity with users and its huge operational costs. Yet OpenAI’s pivot also hints at why AI content’s uncopyrightability may slow mass labor displacement in the creative industries. Licensing content, after all, is expensive. But why pour millions—or, more realistically, if you’re OpenAI and Disney, billions—into an expensive video-generation tool that creates content nobody can turn into commercially viable IP?
Disincentivizing content companies from relying on AI benefits consumers too. Survey after survey shows that audiences value human-made creative works and are skeptical of AI-generated content. The public backlash to Sora’s AI slop also exemplifies this point. Keeping copyright tethered to human authorship ensures a continued supply of the kinds of creative works people actually want to watch, read, and listen to, even if they may be more expensive to make.
[Alex Reisner: Generative AI is challenging a 234-year-old law]
This is not an argument that using AI is unethical or inherently uncreative. Artists should be free to experiment with these tools. Copyright law is mostly about commercial incentives; it’s certainly not a prerequisite to producing content that people find interesting or beautiful. Artists have long worked in media and formats for which copyright offers little or no protection—including conceptual art and improvisational performance—and they still can.
What needs to happen now to ensure that copyright continues to encourage human work? Thaler was the right decision, but the yes-or-no question of whether a fully autonomous AI system can create a copyrightable work is a relatively easy case. The real difficulty lies in drawing a line: How much human involvement is enough to make an AI-generated work copyrightable? The businesses that stand to profit most from a switch from human to AI labor will push to define “human authorship” as loosely as possible, so that a few keystrokes of prompting or a light editorial pass can transform AI output into a copyrighted work. If courts allow that to happen, the structural protections we have described will evaporate. The Copyright Office—the regulatory body that registers much of the copyrightable content produced in the U.S.—has correctly suggested that human prompting alone should not be sufficient to make an AI output copyrightable, but the courts have not yet endorsed this position, and they might come under intense pressure not to.
Courts and regulators also need to impose harsher penalties for misrepresenting AI involvement in copyright registrations and litigation. As AI-generated content becomes harder to distinguish from human-authored work, the temptation to pass off AI-generated content as human-made will grow. The copyright system will protect creators and consumers only if it is backed by meaningful enforcement.
But more than any specific policy recommendation, what we need is to understand this as the key legal question in the fight for protecting human-made creative work. The legal and public conversation about AI and copyright has been consumed by the question of whether training on copyrighted works is infringement. That question will be litigated for years. Even if creators win that battle, their reward might be a onetime payment of a few thousand dollars. But the copyrightability question will determine whether human creators still have jobs in the years to come—and whether the art and entertainment we all consume is made by those humans, or machines that have replaced them.