Update: On June 4, 2023, the Directors Guild of America ("DGA") and the Alliance of Motion Picture and Television Producers ("AMPTP”) reached a tentative agreement, with ratification scheduled for June 6. AMPTP and the DGA have agreed that AI is not a person and therefore cannot be used to replace the duties performed by DGA members. On June 5, 2023, the Screen Actors Guild-American Federation of Television and Radio Artists (“SAG-AFTRA”) announced that 97.91% of the actor’s union voted to authorize a strike if contract negotiations break down with the studios. Stay tuned for further updates.
Over the last few months, artificial intelligence (“AI”) programs, from ChatGPT to Midjourney and other generative AI platforms, have been sparking conversations at the dinner table, within professional offices and educational institutions and at the G7 Summit. Currently, the film and television industry unions are negotiating with the major studios. The goal of the unions is to have standard clauses in their collective bargaining agreements outlining how the industry can use AI.
On May 2, 2023, the Writers Guild of America (“WGA”) began a strike after failing to reach a collective bargaining agreement with the Alliance of Motion Picture and Television Producers (“AMPTP”). The WGA is a Guild representing writers for scripted programming in the film and television industry, while AMPTP represents the major film and television studios and streaming services, such as Walt Disney Studios, CBS, Netflix and Amazon.
Almost 98% of the WGA voted in favor of striking if AMPTP and WGA failed to reach an agreement. Perhaps even more telling is that the WGA is seeing unprecedented support from other labor unions, including the Screen Actors Guild-American Federation of Television and Radio Artists (“SAG-AFTRA”), the Directors Guild of America (“DGA”), the International Alliance of Theatrical Stage Employees, Moving Picture Technicians, Artists and Allied Crafts of the United States (“IATSE”) and the Teamsters. SAG-AFTRA is scheduled to begin contract negotiations with AMPTP on June 7, 2023, and the union has already sent out a vote for strike authorization to members, should negotiations with AMPTP fail. For the WGA, the focus of the negotiations includes fair compensation, residuals, writing credits and other creative rights. However, in addition to the labor and employment concerns typically negotiated in collective bargaining agreements, the WGA and SAG-AFTRA have concerns about studios using generative AI. WGA demands regarding AI include contractual agreements that “AI can’t write or rewrite literary material; can’t be used as source material; and MBA-covered [contract-covered] material can’t be used to train AI.” Some writers want the opportunity to utilize AI in the creative process, without impacting their compensation or credits for scripts. AMPTP rejected WGA’s proposal while acknowledging that technology is advancing, but offered only to hold annual meetings to discuss the changing technology. AMPTP has not released any official statements, but writers are concerned that if existing scripts are used to train AI, then studios will use AI to write complete scripts. If this happens, writers may be brought on to productions as consultants to edit AI written scripts, greatly reducing their compensation and creative control.
AI and Copyright Questions
The use of AI as a tool and not a replacement for writers creates a number of interesting legal questions, particularly under copyright law. Currently, material created by AI cannot be copyrighted because under §102(a) of the United States Copyright Act, in order for a work to be eligible for copyright protection, it must be an “original work of authorship fixed in any tangible medium of expression.” AI generated work is machine made and thus is not considered a work of authorship.
The Copyright Office has released The United States Copyright Office, Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence (2023) on works containing material generated by AI. Typically, in AI-generated works, the traditional elements of authorship are done by the AI itself. Even when the AI allows the human user to input instructions that generate the material, because the AI is “exercis[ing] ultimate creative control over how such systems interpret prompts and generate material,” the human user cannot be credited as the author. However, in the guidance, the Copyright Office has stated that there are times when AI-generated material can be copyrightable:
In other cases, however, a work containing AI-generated material will also contain sufficient human authorship to support a copyright claim. For example, a human may select or arrange AI-generated material in a sufficiently creative way that “the resulting work as a whole constitutes an original work of authorship.” Or an artist may modify material originally generated by AI technology to such a degree that the modifications meet the standard for copyright protection. In these cases, copyright will only protect the human-authored aspects of the work, which are “independent of” and do “not affect” the copyright status of the AI-generated material itself.
The Copyright Office has hosted multiple events to seek input from relevant constituencies and to provide additional guidance in advance of releasing a notice of inquiry to be published in the Federal Register later in 2023.
At this point it is unclear if scripts generated by AI and then edited by writers and studios would have sufficient human creative control to be considered copyrightable. Recently, in response to a copyright application for Zarya of the Dawn, a comic book featuring AI-generated artwork by Kristina Kashtanova, the Copyright Office granted copyright protection to the work’s text, as well as the selection and arrangement of images and text, because Kashtanova was the author. However, the Copyright Office determined that the images in the work, which were created using an AI system called Midjourney, were not protectable under copyright because they could not be considered “original works of authorship.”
This standard was articulated in 1884 by the Supreme Court in Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53, 61 (1884), where the Court stated that the author of a copyrighted work is the one “who has actually formed the picture,” who acts as “the inventive or mastermind.” Midjourney, like many generative AI systems, generates images based on the prompts of human users. The human instructions may influence the work created by the AI, but it does not decide the fixed outcome. Furthermore, users don’t have control of the final image generated. Thus, the Copyright Office stated that images rendered by Midjourney are not protectable under copyright.
However, when creators can control the final image such that it amounts to the artist’s “own original mental conception, to which [they] gave visible form” (Burrow-Giles, 111 U.S. at 60), then the image is considered copyrightable. AI script generators are distinguishable from AI image generators because the studio using the AI-generated base script can easily adjust the text so that it reflects their “original mental conception,” making the studio the “mastermind” of the final work. In Zarya of the Dawn, Ms. Kashtanova made some edits to the final images used in the comic, but the Copyright Office believed that the edits were de minimis and did not “contain a sufficient amount of original authorship.” The Copyright Office said that “[t]o the extent that Ms. Kashtanova made substantive edits to an intermediate image generated by Midjourney, those edits could provide human authorship and would not be excluded from the new registration certificate.” Thus, should AMPTP studios use AI to create a basic script and then hire writers as consultants to edit the script, the final script could be eligible for copyright protection.
Threats to Other Creatives from Rapidly Advancing Technology
Studios and AMPTP believe that at this point AI is not capable of creating production-worthy scripts, and thus should not be part of the current contract negotiations. However, the technology is advancing rapidly, and writers legitimately fear being replaced. Furthermore, writers are not the only members of the creative industry who worry about AI.
Technology has evolved to a point where it can create “deepfakes,” that make it appear as if a person is doing something they have never done. A deepfake is a type of artificial media that uses AI to alter faces in existing images and videos to appear as other individuals. Deepfakes can be created by using sample video footage of a person, which can generate a realistic looking and sounding avatar, or from a still image which creates a less realistic avatar with limited facial expressions.
Deepfakes of prominent celebrities, such as Tom Cruise and Harry Styles, have gone viral on social media, raising questions about right of publicity, satire and parody as well as ethical concerns. This technology can also be used by studios to place actors into films, as demonstrated in recent films such as Furious 7, that used AI and computer-generated images ("CGI") to allow deceased actors to “appear” in additional scenes.
Some argue that allowing actors or estates to sell their right of publicity or simulation rights to film studios may provide actors and their heirs more revenue opportunities, because an actor’s name and likeness could be used without any additional work on their part. Others worry that AI-generated performances remove the humanity and heart from the artistic work. Furthermore, a performer or their representative who does not have the final say on the AI-generated performance or the type of projects in which the production company is permitted to use their AI suffers a critical loss of their agency over their image and reputation. While actors normally do not get a say in the final cut of a project, they should always have the opportunity to decide if they want to appear in a project at all. As simulated performances become more mainstream, SAG-AFTRA has already stated that producers will have to bargain with the union for “terms and conditions involving rights to digitally simulate a performer to create a new performance.”
AI is not only able to create new performances, but it can also be used to edit existing performances. For example, AI is being used in an upcoming film, Here, directed by Robert Zemeckis, to de-age Tom Hanks and Robin Wright. Previous industry standards for making actors look younger involved talented hair and makeup artists, specific lighting choices, digitally mapping a performance using dots and special cameras, or casting a younger actor to play the younger version of a character. Increasingly AI is able to do the job of these professionals. In the 2019 film, The Irishman, a combination of makeup, lighting, special cameras and CGI made 76-year-old Robert De Niro appear to be the 20-year-old character. Here is using a significantly less expensive Metaphysic AI to de-age actors in real time, without the need for post-production special effects.
When AI software is used by filmmakers to enhance the visual aspect of an actor's existing performance, as seen in Here, the AI is acting in a way similar to photoshop or other standard CGI techniques. With CGI techniques, actors still make acting choices and then in post-production, visual effects artists add the special effects to the performance. Metaphysic AI takes out the post-production step and alters the film in real time, like a digital camera filter on a phone app. Some actors already have clauses in their contracts limiting how their likeness can be photoshopped. With the rise of AI, SAG-AFTRA wants actors to have the opportunity to have specific clauses in their contracts limiting how AI can be used in the production or post-production of a film.
AI and the Music Industry
AI is also a concern for the music industry where AI can replicate an artist’s voice and make it sound as if that artist is performing a song. Recently, an AI-created song went viral because it sounded like it was being performed by Drake and The Weekend, when in actuality neither artist participated in the performance. Like the deepfakes in film or television, this use of AI highlights issues of right of publicity, copyright, credit and compensation in addition to raising ethical concerns.
The technology has become so widely available that anyone can make it sound as though an artist is performing a song that they have never performed. These songs can go viral on social media and affect the artist’s image and reputation even if the third-party content creator is not profiting from their AI creation.
The National Music Publishers Association (“NMPA”) has written a letter to Congress, urging it to “consider the impact of AI technologies on the creative industries in the U.S.” because such technologies impose a “threat to human creativity.” NMPA requested that Congress regulate how AI uses copyrighted materials as training data and to be transparent about what works have been used in the training data of the AI. NMPA wants third parties to be required to have the artists’ prior consent for their works to be used in training sets, as well as requiring the artists to receive credit and fair compensation.
The WGA strike is likely to be the first of many contract disputes involving AI in the entertainment industry and beyond. As noted above, support of the WGA strike from the other unions in the entertainment industry is unprecedented; professional sports unions, writers unions from across the globe, government officials, and other labor unions have also expressed their support of the WGA. Currently, IATSE and Teamsters are not crossing picket lines, halting production on shows.
POSTSCRIPT: The AI debate harkens back to work that Arnie Lutzker did in the 1980s and 1990s on behalf of the Directors Guild of America, its Artists Rights Foundation, and other creative unions in Hollywood opposed to colorization of classic black and white movies and TV shows. In the mid-1980s, Ted Turner bought the MGM film library, including movie masterpieces like Citizen Kane and It’s a Wonderful Life, and famously announced that since kids won’t watch black and white movies, he was going to digitally colorize them regardless of the wishes of the creators (directors, screenwriters, cinematographers and others in the industry).
At the time, Arnie counseled the DGA and the other creative guilds to join the Congressional debate over the ascension of the U.S. to the Berne Convention and the application of Article 6bis to U.S. copyright law. Berne, the granddaddy of international copyright treaties, was inspired by 19th century legends including Victor Hugo, and Article 6bis guaranteed protection of the “moral rights” (the French droit moral) of creators. In essence, moral rights are the personal and reputational rights of authors that adhere to their creations. Most notably and independent of economic interests, droit moral assures that the person who created a work (script, book, film, photo, etc.) has the right to claim authorship of the work and to object to any distortion, mutilation or modification of the work which would be prejudicial to his or her honor or reputation. Just as AI challenges the definition of authorship by relying on existing works and images to generate novel content, colorization took existing works and transformed them into something new.
In future insights, Arnie and the L&L team will have more to say about lessons from the Berne debate, and additional developments regarding AI, the writers’ strike and the upcoming negotiations with the other Hollywood unions. We encourage you to stay tuned!