Near the height of animosity in the writers’ strike in 2023, the Alliance of Motion Picture and Television Producers outlined an offer that captured its members’ anxieties in the adoption of AI tools in the production pipeline. In what it hoped would be perceived as a concession, it proposed that scribes’ compensation and rights wouldn’t be undercut if any part of a script is based on AI-generated material.
The Writers Guild of America balked at the offer. It knew it had some leverage as far as negotiating points involving AI. That’s because works solely created by the tech aren’t copyrightable. To be granted protection, a human needs to rewrite any AI-produced script. And by keeping AI on the table, the guild figured that the studios were looking to capitalize on the intellectual property rights around works created by the tools. “If a human touches material created by generative AI, then the typical copyright protections will kick in,” a source close to the AMPTP told The Hollywood Reporter at the time.
Almost two years later, that proposal continues to reflect the tenuous nature of copyrightability for movies and TV shows in which AI is utilized. Under the U.S. Copyright Office’s current guidelines, refined in a report issued on Wednesday, there’s copyright protection if the author “selects and arranges” AI-generated elements in a manner that’s sufficiently creative. AI proponents and naysayers were both quick to declare the findings a win for their respective sides. But missing from the report is clarity on what rises to an adequate level of creativity.
“Where that creativity is expressed through the use of AI systems, it continues to enjoy protection,” said Shira Perlmutter, Register of Copyrights and Director of the U.S. Copyright Office, in a statement.
Consider the use of AI in Brady Corbet’s Academy Award-nominated The Brutalist. In that film, AI tools were used to tweak the dialogue of the movie’s leads, Adrien Brody and Felicity Jones, and as inspiration for a series of buildings supposedly designed by Brody’s character, the fictional architect László Tóth (the final images depicted in the movie were hand-drawn). Corbet, in an interview with THR on Monday, defended the decision, saying that the tech was only utilized to “refine certain vowels and letters for accuracy.”
Ethics aside, do those scenes in the movie enjoy copyright protection (they likely do)? The question boils down to whether AI, as the copyright office’s report characterized it, was used to “assist rather than stand in for human creativity.” The uncertainty has forced studios, production entities and distributors to navigate what they consider an overly rigid intellectual property scheme that doesn’t take into account the entertainment industry’s longtime use of AI tools in the production pipeline.
Under the current copyright registration process, any AI use must be disclaimed for an examination of how the tech was incorporated into the final work and whether it should enjoy protection. This includes utilization of AI as a production and postproduction tool in the hands of human creators, including rotoscoping, color correcting, detail sharpening, deblurring and removing unwanted objects. Amid a downturn in production that’s left many in Hollywood without jobs, AI has prompted fierce backlash for further eroding available work in the industry, but few would push back against the use of tools in the production pipeline that automate and speed up certain monotonous tasks.
A major studio production involves dozens of individuals working on writing, art, sound, costume, editing and special effects, among other jobs. Many of them contribute to the creativity of the finished project. And some use AI, complicating the assurance of copyrightability for a movie or TV show.
The Motion Picture Association has said that the fact some creators produced parts of a movie with the help of AI shouldn’t render those portions uncopyrightable. Take a hypothetical example of a superhero movie. The movie may be copyrighted, but what about scenes involving AI-assisted visual effects depicting a space battle? At the forefront of studios’ concerns: the protection of their rights if the underlying characters and scripts are protectable but not certain portions that incorporate material generated by AI. A copyright carveout for certain scenes, the MPA has said, isn’t tenable. It’s argued in favor of not having to comply with the copyright office’s registration requirements since applying them would have “significant, negative real-world consequences.”
Provisions in U.S. intellectual property law barring the copyrightability of AI-generated material severely hampers the commercial viability of productions that utilize the technology since they would enter the public domain. Several studios forbid the use of AI in writers’ room and have some scribes sign certificates of authenticity attesting that they wrote scripts themselves.
“Contracts say you need to ask permission of studios, and a lot of studios’ policies is that it’s simply not allowed,” said showrunner and writer Mark Goffman (Bull, Limitless, The West Wing) at AI on the Lot, a conference AI in the entertainment industry last year.
This is in line with some distributors having production entities for projects they buy attest that generative AI wasn’t used in the moviemaking process.
It remains to be seen whether regulators move toward a copyright scheme that’s more receptive to protection for works in which AI was utilized. Complicating the conversation is the possibility of courts ruling that the training of AI systems on copyrighted works constitute infringement, though most visual effects workers that have adopted the technology use open source models that don’t pose that risk.
Consistent with earlier guidance, the copyright office clarified that current technology doesn’t afford users enough control of the ultimate output for the use to be considered sufficiently creative.
“In theory, AI systems could someday allow users to exert so much control over how their expression is reflected in an output that the system’s contribution would become rote or mechanical,” the report states. “The evidence as to the operation of today’s AI systems indicates that this is not currently the case.”
