Back to selection
What Filmmakers Should Know about AI Copyright Restrictions
(Photo: Gearstd/Shutterstock.com) Uttering the words “artificial intelligence” in Hollywood right now elicits something of a Chicken Little response. From major studio IP franchises to small independent documentaries, there is no corner of the entertainment industry that artificial intelligence does not (or will not) touch and, with so much uncertainty surrounding the legality of AI, many industry stakeholders have taken to wringing their hands and proclaiming that the sky is falling. Despite artificial intelligence’s novelty, however, many of the issues surrounding its legality can be addressed by pre-existing copyright and First Amendment principles.
Copyright Guidelines Governing “Human Authorship”
For example, let’s examine the copyrightability of works generated by artificial intelligence. Recently, the United States Copyright Office and the United States District Court for the District of Columbia made headlines by determining that AI-generated works are not protectable by copyright because they are not a product of “human authorship.” The fundamental principle that human authorship is a prerequisite for copyright protection dates back to the founding of the United States. According to the U.S. Constitution, Congress has the power “to promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries.” In the seminal 1884 case, Burrow-Giles Lithographic Co. v. Sarony, the Supreme Court held that an “author” is “he to whom anything owns its origin; originator; maker; one who completes a work of science or literature.” Various courts have since interpreted this definition of author to include an implied element of humanity. In the 1997 case, Urantia Foundation v. Maaherra, the Ninth Circuit declined to recognize copyright protection for text within a book that claimed to embody the words of celestial beings. According to the court, if the origin of the text were true, the text would have been the product of divine creativity, rather than human creativity, and therefore uncopyrightable. Likewise, in the 2011 case Kelley v. Chicago Park District, the Seventh Circuit declined to recognize copyright protection in a garden because, although humans were responsible for cultivating the garden, the garden ultimately owed its form to the forces of nature, rather than human intervention. The 2018 case Naruto v. Slater also addressed this issue when the Ninth Circuit declined to recognize copyright protection for a “selfie” photograph taken by a monkey in an Indonesian wildlife reserve.
How the Courts Have Historically Viewed AI-Generated Works
Based on this case law, it should be no surprise that a court has held that works generated by artificial intelligence are similarly not protectable by copyright. In the 2023 case Thaler v. Perlmutter, a computer scientist, Dr. Thaler, created an artificial intelligence system called “Creativity Machine” that generated a piece of visual art that he titled “A Recent Entrance to Paradise.” After the Copyright Office denied Thaler’s application to register the work for copyright protection, he filed suit. The court acknowledged that copyright must adapt with the times and stated that “Underlying that adaptability… has been a consistent understanding that human creativity is the sine qua non at the core of copyrightability, even as that human creativity is channeled through new tools or into new media.” Based upon this, the court upheld the Copyright Office’s decision to deny copyright protection to the scientist’s work.
But does this mean that a film that includes artificial intelligence-generated works will not be protected by copyright? Not necessarily.
Although the Copyright Office will not grant copyright protection to works generated solely by non-human authors, it will grant copyright protection to a human’s creative arrangement and selection of those works, or to elements of an overall work that was authored by a human. For example, although the Ninth Circuit declined to extend copyright protection to the text within the book in Urantia Foundation v. Maaherra, the court did find that the book as a whole could be protected by copyright because human creativity was involved in the organization and arrangement of the divinely originated text within the book. More recently, the Copyright Office affirmed that the selection and arrangement of works generated by artificial intelligence could be protected by copyright when it addressed the copyrightability of a comic book entitled Zarya and the Dawn. While the Copyright Office initially approved the book’s copyright application and granted copyright protection in 2022, it subsequently reopened the application after learning that the images within the book were created by Midjourney, a generative AI model. After investigating, the Office refused to grant protection to the images themselves and held that the creator’s input of a text prompt into the model was insufficient to establish human authorship over the images. However, the Office did grant copyright protection to the wholly original text within the book, as well as the creative selection, coordination, and arrangement of the text alongside the AI-generated images.
The Future of Film: Is AI Here to Stay?
Based on this precedent, it is clear that films containing works generated by artificial intelligence can still maintain copyright protection for the film as a whole. Indeed, according to the Copyright Office’s guidance, filmmakers registering works that contain AI-generated material must merely disclose the inclusion of all such material within the copyright application. The Office will then exclude the material from the registration in the same way that pre-existing materials such as photographs, video footage, and music have always been excluded from new works.
Previously established legal principles can also be applied to address the inclusion of artificially generated “digital replicas” in films. To address concerns about this technology, state legislatures have turned their attention to pre-existing right of publicity statutes.
AI and the Right of Publicity
Stated generally, the right of publicity is the inherent right of every individual to control the commercial use of their identity by prohibiting the use of their name, image, or voice without their consent. Although there has long been discussion of a potential federal right of publicity, rights of publicity are currently governed by individual states and the relevant laws of each state vary. Not all states have right of publicity legislation, but such legislation currently exists in 33 states. Some states that do not have right of publicity statutes recognize rights of publicity under common law.
Notably, right of publicity legislation is limited to the commercial exploitation of an individual’s likeness without their consent. For example, in the 1988 case, Midler v. Ford Motor Co., the Ninth Circuit found that Ford violated famous singer and actress Bette Midler’s right of publicity by hiring Midler’s backup singer to imitate her singing voice in their advertisements for the Lincoln Mercury. Four years later, in White v. Samsung Electronics America, Inc., the Ninth Circuit found that Samsung similarly may have run afoul of Wheel of Fortune host Vanna White’s right of publicity by including a robot dressed in a wig, gown, and jewelry in front of a Wheel of Fortune game show set in a series of video-cassette recorder advertisements.
How has AI implicated right of publicity law? Mainly in the area of so-called “digital replicas” generated by AI models to imitate a person’s likeness or voice. Due to pressure from SAG-AFTRA, estates of celebrities, and other groups, recently some states have amended their right of publicity laws to prohibit or restrict the use of digital replicas. To date, only New York, Louisiana, Tennessee, and California have such laws, but more are near certain to follow. Just like pre-existing right of publicity laws, each state’s digital replica legislation has specific nuances that impact its practical applicability. The applicability of protection, for example, can depend on whether the replicated individual is deceased or alive. It can also depend on whether the individual is (or was) a professional performer or an average citizen.
Importantly, all states that have enacted digital replica legislation have ensured that there are carve-outs for expressive works to ensure that they do not run afoul of the First Amendment. This is in line with well-established precedent that expressive works, which embody freedom of expression, are exempt from right of publicity laws.
In de Havilland v. FX Networks, Inc., the California Court of Appeals rejected famed Hollywood actress Olivia de Havilland’s right of publicity claim based upon her depiction in a television docudrama entitled Feud about the feud between Bette Davis and Joan Crawford. According to the court, the series was protected by the First Amendment and, therefore, neither Olivia de Havilland nor any other individual had a “legal right to control dictate, approve, disapprove, or veto the creator’s portrayal of actual people.” Courts have also affirmed that right of publicity protection has no place within documentary filmmaking. In 2019, the Southern District of New York dismissed singer Bobby Brown’s claims that the documentary Whitney: Can I Be Me? violated his right of publicity and that of his late daughter, Bobbi Kristina. As the court explained, “a right of publicity cause of action may not be maintained against expressive works, whether factual or fictional because without such First Amendment protection, reports and commentaries on the thoughts and conduct of public and prominent persons’ would be subject to censorship under the guise of preventing the dissipation of the publicity value of a person’s identity.”
Statutory and judicial carve-outs, in both right of publicity and digital replica legislation, are necessary to ensure the public’s freedom of speech and expression under the First Amendment. Without such exemptions, any biography of a person, or any film about a historical event, for example, would run afoul of the right of publicity law and, as a result, violate free speech or be subject to censorship by the persons depicted or their estates.
When considering the “scary new world” of digital replicas, it is helpful to remember that the replication of individuals in television and film is not unchartered territory. While not utilizing artificial intelligence per se, filmmakers have long used practical effects, such as actors, makeup and costuming, to replicate real-life individuals. Take, for example, Charlize Theron’s portrayal of serial killer Aileen Wuornos in Monster or Gary Oldman’s portrayal of British Prime Minister Winston Churchill in Darkest Hour (both of which were honored with Academy Awards). Filmmakers have also used digital technology to recreate historic events, such as in Forrest Gump, where Tom Hank’s beloved character inspires John Lennon to write his hit song “Imagine” and meets President Richard Nixon before unknowingly exposing the “Watergate” scandal. These traditional filmmaking techniques do not violate rights of publicity, and the AI version of the same should not either.
Put simply, while the advent of artificial intelligence in film and television is exciting, it is not so groundbreaking as to necessitate the creation of new legal principles. Legal concerns can be adequately addressed by pre-existing law that has successfully protected artists’ creations and First Amendment rights for hundreds of years.
