Exploring AI Matters
Our mission is to help the policy community understand the breadth and richness of AI and the potential for such technologies, wisely applied, to augment all sorts of human endeavors.
Some AI tools are able to assist humans in performing tasks faster, more accurately, or more efficiently. Some, however, are inaccurate and unreliable. Who or what we hold accountable for these flaws, and what incentives we do or do not create for their correction will influence AI’s hand in how we work.
In this series we will refine, sharpen, and clarify your understanding of AI.
Exploring AI Matters
Episode 21 - Hollywood Adventures in GenAI
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
The entertainment industry in the US, with revenues in 2022 of about 8.5 billion dollars, is only three hundredths of one percent of the 23.3 trillion dollar US economy. However, entertainment is far more important to society as a whole than the industry’s financials might suggest.
In 2022 when large language models like OpenAI’s ChatGPT and Google’s Gemini began to show signs of prodigious capabilities in the entertainment domain, studios began to experiment with using AI in lieu of actors and writers. In response to the refusal by the studios to negotiate guardrails for the use of AI, the unions representing the writers and actors went out on strike and forced negotiations.
The agreements concerning the status of generative AI that were instrumental in settling the strikes are a tremendously important step in establishing governance of AI in operational practice, at least for one industry. These agreements may hold important lessons for other industries.
We are fortunate to have as our guest for this episode of Exploring AI Matters Louise Nemschoff, who practices entertainment and copyright law in Los Angeles. She represents a wide variety of parties in both domestic and international transactions.
Welcome to Exploring AI Matters. This podcast series, previously known as Mind the Gap, Dialogues on Artificial Intelligence, will continue to appear in the ABA series to the extent that, in addition, all of the episodes, old and new, will now appear under our new podcast name, Exploring AI Matters. Thank you. However, entertainment is far more important to society as a whole than the industry's financials might suggest. Entertainment, comprised of movies, TV, live theater, books, magazines, recorded and live music, and games, provides important spice to our lives, providing cultural experiences, helping enrich our relationships with people in our lives, and shaping our view of ourselves, our society, and our world. In 2022, when large language models like OpenAI's Chat GPT and Google's Gemini began to show signs of prodigious capabilities in the entertainment domain, studios began to experiment with using AI in lieu of actors and writers. In response to the refusal by the studios to negotiate guardrails for the use of AI, the unions representing the writers and actors went out on strike and forced negotiations. The agreements concerning the status of generative AI that were instrumental in setting the strikes are a tremendously important step in establishing governance of AI in operational practice, at least for one industry. These agreements may hold important lessons for other industries. Welcome to Mind the Gap Dialogues on Artificial Intelligence, Louis Nemshaw. I'm Amma Adams, a national security lawyer.
SPEAKER_04And I'm Mark Donner, a computer scientist. We are your hosts for this episode of Mind the Gap Dialogues on Artificial Intelligence. In addition, we have two more hosts.
SPEAKER_05Hello, I'm Roland Trope, a national security lawyer.
SPEAKER_03And Charles Palmer, a computer scientist.
SPEAKER_00Each episode will be led by two of us, with the others adding impromptu questions and comments as the spirit moves them. We are fortunate to have as our guest for this episode Louise Nemshaw, who practices entertainment and copyright law in Los Angeles. She represents a wide variety of parties in both domestic and international transactions. In addition to representing clients, Ms. Nemshaw mediates entertainment industry and intellectual property disputes and serves on the arbitration panel of the International Center for Dispute Resolution for the Independent Film and Television Alliance. An active member of the Los Angeles County Bar Association Intellectual Property and Entertainment Law Section, she served as its 1993 to 1994 chair and completed a two-year term on the Board of Trustees of the Los Angeles County Bar Association.
SPEAKER_04Welcome to Mind the Gap, Louise. Thank you for your time. Can you help us understand the structure of the movie and TV industry, which we understand is only part of the entertainment industry?
SPEAKER_01Well, thank you all for inviting me. To begin with, Hollywood has, of course, a corporate structure like most industries. But when it comes to the actual production of films and episodic programming for television and video on demand, the industry is actually more of a gig economy. So writers, directors, actors, cinematographers, editors generally work on a project-by-project basis, like a particular film or a series for TV or for video on demand for streaming platforms. This means they could be working just a few weeks, they could be working for several months, and with options to renew their services, the work could actually stretch out over several years. Some but not all of this talent is represented by a union, not a single union, a group, a group of unions. These are sometimes known as the guilds. This includes writers, directors, and actors, among others. As a result of this, however, there are both union and non-union productions being made because not all talent belongs to a union.
SPEAKER_04Can you help us understand the key terms for roles and responsibilities that historically operated in the movie and TV industries? Specifically the ones that we've encountered in this conversation is Ayatse, the Studios, MPA, AMPTP, SAGAFTRA, and WGA.
SPEAKER_01Okay, well, there are a number of unions, including the Writers Guild, which is sometimes referred to as the WGA. There's the Director's Guild, which is often referred to as the DGA, the Screen Actors Guild, which is called SAG AFTRA, and then there are many local unions of the larger overarching international alliance of theatrical stage employees known as AYATSI, IAT S E. Its initials. To help everyone with these acronyms and nicknames, we've compiled a short glossary of terms, at least the terms I'm going to be using most in this podcast. Now, on the other side of the fence, the unions negotiate with what's called a multi-employer collective bargaining association. In Hollywood, this is the Alliance of Motion Picture and Television Producers, also known as the AMTTP, sometimes also called the Alliance. In this regard, it's like the construction or some other industries that allow for centralized collective bargaining negotiations and allow the talent of the workers to work for different production companies under the same common collective bargaining agreements. The AMPTP represents over 350 production companies who are signatories to the collective bargaining agreement. That means they signed on and agreed to accept and abide by its terms. However, the AMPTP is often referred to as the studios for shorthand. This dates from a time when the traditional motion picture studios were the primary members of the AMPTP and were certainly the lead negotiators in collective bargaining. Today, the AMPTP members include broadcast and cable TV stations, many of which are now owned by the studios, and also the streaming platforms like Apple, Amazon, and Netflix. When successful, the negotiations result in a collective bargaining agreement. In Hollywood, these well, collective bargaining agreements are for a specific period of time. In Hollywood, the term of each agreement usually runs about two or three years, and their expiration is staggered amongst the various unions, so they don't all expire at exactly the same time, but they expire often within months of each other. In 2023 and 2024, many of these collective bargaining agreements have come up for renegotiation. The contracts for the DGA, the WGA, and SAGAFTRA all expired in late 2022 or early 2023.
SPEAKER_04So can you give us an overview of how concerns over AI caused the dispute between the writers and the studios that led to a 148-day strike?
SPEAKER_01Well, AI had been on Hollywood's radar for some time. Hollywood was already familiar with computerized production tools like DGI, computer generated images, especially for special effects, but also for making a number of corrections in images. They use motion capture photography and also a number of computerized editing tools, like the tools with the Avid editor. As you noted, OpenAI released ChatGPT in November of 2022, and the buzz about generative AI grew exponentially in 2023. As with most new technologies, concerns arose about the use of generative AI that it might lead to major industry changes and job loss. The first of the guild contracts to expire was the DGA. At that time, the DGA concluded that threats to its membership, namely directors, assistant directors, and unit production managers, was at least two or three years away and could be covered in a later round of negotiations. So basically, they just agreed rather quickly to new terms, including the only AI provision, which was that the parties would continue to have discussions on the subject of AI and would address it at a later time. The WGA, however, came to a different conclusion. They knew that the ability of generative AI to generate text raised a number of immediate concerns for their members that might result in their possible replacement or displacement in the screenwriting process.
SPEAKER_00So go ahead. No, I was just going to ask if you could drill down a little bit more on that for us. You noted that for the WGA, they certainly saw the threat looming from generative AI, sort of the use and status of generative AI as something much more imminent than perhaps some of their other counterparts, so to speak. And so would I think be of interest to really understand sort of what some of those threat scenarios that the writer's guild was, writer's guild was focusing on that helped drive the decision to strike.
SPEAKER_01Well, let me start with a continuum. On the one hand, the directors and the DGA decided there was no imminent threat. Maybe in a few years or even a few months, but not immediately. SAG, on the other hand, SAG AFTRA and their members were concerned that the recorded images of actors would be used to create replicas of their images, which could then be programmed to create entirely new performances without having to actually hire most actors. I think in general, this is a topic for another podcast, but it is obviously an existential threat for most actors. The writers fall somewhere in the middle. I mean, certainly AI is not yet able to write jokes, and its writing so far doesn't do much to generate emotions in the readers. But the writers nonetheless had about five primary concerns. First, they were concerned that writers would be required to rewrite a script generated by AI. This was a problem because the payment structure for writers, particularly for films, but also for episodic theories, was and is based on the stages of work done on the development of a script. The development of a script usually begins with a story, which may be in the form of an original treatment or some other things, which I'll mention shortly. But a writer is paid a certain amount to prepare a treatment or a story summary. The writer's paid more to write a first draft screenplay than a lesser amount to do a rewrite. And even smaller amounts are provided for doing what's called a polish of a script. Maybe polishing up the jokes, maybe polishing up the dialogue, but something less than a rewrite. Before actually shooting a film, in fact, a screenplay may go through multiple treatments, multiple script drafts, many revisions, and many polishes at the hands of multiple writers. To the extent that writers would be delegated to rewriting or polishing the output of generative AI, their opportunities to work and their compensation for that work would be reduced, being relegated to lesser levels of compensation on the WGA rate card. Also, and this is important to note a writer's compensation on his or her next project may be determined as custom and practice in the industry by what they were paid on their last gig. So this has long-term the relegation to a lower level of payment has long-term effects on a writer's career earning potential. The use of AI to generate a screenplay would also affect or eliminate the credits that might be awarded to a writer on the production. Credits are, of course, what you see in the roles at the beginning or the end of films. Most people outside the entertainment industry ignore them. But they're very important to people in the industry. With writers and with everyone, credit is used to determine the person's relative standing in the Hollywood hierarchy. Were you credited on a hit film, or was your contribution not credited because it was considered so minor? Credit also determines who is entitled to many types of post-release payments. These are subsequent payments after the film is released, which are dependent on where and in what media the film is released and how financially well the film does. These include things like residuals and profit participations, which can vary in value from zero to many, many hundreds of thousands or even millions of dollars. The second concern of the guild was that AI would generate source material for a screenplay that a writer would then be asked to prepare. So they would be given, the writer would be given a treatment or other narrative summary of the story, a short story, for example, or a novel or a stage play, and be told to adapt it for a screenplay. Adapting such material again could result in lower compensation for the writer, as well as impacting credit. It also parenthetically determines which writing Oscar a writer might be eligible to receive, whether it's for original screenplay or adapted screenplay. Also, if a writer were forced to work off generative AI created source materials, writers might not be eligible to receive certain reserved rights, known as separated rights, which are guaranteed for writers who create original screenplays, not adapted screenplays. These uh separated rights include things like the right to publish the script itself, as Brenton Tarantino tried to do when he tried to sell an NFT of his script in uh one of his major films, and ended up getting sued by the production company, the case settled, by the way. So we never know, we'll never know at this point whether he was entitled to do that under his separated rights. Separated rights also include the right to produce a live stage version, which can be very profitable. And also, from a writer's perspective, they would receive an automatic paid rewrite. This means additional guaranteed compensation and more input into the final script, which may help the writer in getting credit for his or her work. And in general, it avoids a writer getting cut out of a project at a relatively early stage. The third writer's concern was that generative AI could be used to rewrite, edit, or polish a writer's screenplay, thus depriving the writer of additional compensation that they would otherwise receive for either doing the rewrites or polishes themselves, or which would be given to another writer hired to do that work. This also has an impact on credit. Fourth, the writers were concerned that they would be required by their employers, that is, the studios and signatory production companies, to use generative AI as a condition of their employment in the course of doing their development work. This could undermine the writer's creative process, be used to negotiate or apply lower rates of compensation, and impact a writer's credit. Fifth, writers were concerned about who owned the right to authorize AI large language models to be trained on their pre-existing and future works. This requires a little background explanation. The copyright in the final screenplay, and much, if not all, of the earlier development writing work, would in most cases be owned by the studios as works made for hire. However, when a writer creates what's called a spec script, one that is created by a writer on his or her own initiative and not under any employment contract, then the writer under copyright law is the author and initial owner of the copyright in that script. The employer will invariably include in its writer contracts provisions saying that all rights are assigned to the employer if it's not a work for hire. But as you may be aware from one of your earlier podcasts on copyright and AI, a writer and his heirs retain certain recapture rights, even if the script is assigned to another party. These are known as termination of transfer rights. The studios usually insist on including various provisions in their writer agreements, which either grant those termination of transfer rights to them or at least try to undermine them. For example, a right of first refusal, giving the studio an opportunity to make it difficult for the writer to sell those termination of transfer rights or the recovered rights in the script to anybody else. However, to my knowledge, these provisions have not yet been tested in court. And it will be interesting to see how that turns out because the purpose of the termination of transfer rights was to restore rights to creators. And there is some language that suggests that the parties cannot contract around those provisions. So those were the five major concerns that the WGA had, and the reasons for their concerns, particularly their impact on compensation, credit, and copyright ownership.
SPEAKER_05It strikes me that those are remarkably focused. Concerns. In many industries, when you hear talk about the threat of AI, it's always some sort of generalized, obscure threat that will wipe out large numbers of jobs or reduce opportunities for people to get certain kinds of work. Is there something that caused this particular industry to be so focused in identifying what their concerns were that they actually, from what you're describing, seems to have made them probably more negotiable than just a vague threat?
SPEAKER_01Well, the writers guild and the studios have for many years signed and worked under what's known as the basic agreement. And this covers everything from rates of compensation, ownership, credit, arbitration of disputes, and there are many arbitrations relating to credit disputes, how credits are determined when there are multiple writers, and a variety of other topics. And over the years, they've developed structures in all these different areas, like different compensation levels for different levels of work, like guidelines on how credits are to be determined, like treatment of original material, an original screenplay versus an adapted screenplay, and the application of separated rights to the original material. So they had this pre-existing structure, and they were able to negotiate and identify concerns within that structure, which made it possible for them to negotiate an agreement once the studios were prepared to sit down and discuss it. And so I think it's very important in other industries, unionized or not, that the party who is concerned about AI try to prepare carefully and think carefully about how AI could specifically affect them. Not just, oh, I'm going to be thrown out of a job. Sure, that's going to happen in some cases. I mean, in many ways, AI is to white-collar work, what robotics was to blue-collar work. Neither will completely eliminate jobs, but it will change the nature of those jobs dramatically. So if you're trying either as a union or a non-union employee to enter into contracts that may be affected, where your work may be affected by AI, it's important to think about the specific ways in which your work will be affected and what tools you can use and what structures you can use to mitigate them.
SPEAKER_00So stepping back just a bit, right, we kind of walk through some of the traditional issues that might come up in a sort of negotiation between the guilds and the studios. And then we have the overlay of AI, you know, just generative AI and the threats that you just walked us through, those five concerns or threats. It's complicated. These are not easy issues to kind of mirror together and sort of find a path forward. And in your view, what made the WGA and SAG AFTRA so effective or successful in bringing the studios to the table to grapple with these issues and try to drive sort of a framework going forward?
SPEAKER_01Well, Ama, you're absolutely right. This is quite a complex set of contracts. I mean, the closest thing I can analogize it to is the tax laws. And they require a good deal of familiarity and expertise to even work your way through the language of these union basic agreements, which set minimum terms of employment. But why were the unions so effective? Well, there are a number of factors, but certainly the most important was that both writers and actors were on strike for much of the same time. The writers were in strike for 148 days, their contract expired first, they went out on strike first. Then the SAG after contract expired, and the actors went on strike for something like 114 days. This the directors who had their own agreement were not on strike, but there was nothing for the directors to do because there were no scripts for them to direct and no actors to be directed. So this effectively shut down production on everything except perhaps animation and reality TV for months. Like many other industries, the entertainment industry has a release schedule. It takes six months to a year to produce from start of principal photography to delivery of a final print, six months to a year, sometimes two years to get a film completed. Television series, individual episodes are, of course, faster than that. But there's this time lag between when they start on a work and when they can release it. And plus, they want to do promotion and marketing before it's released to generate the biggest audiences. So this explains why there have been relatively few major motion pictures released in the first quarter of 2024. They weren't written, they weren't shot. And in fact, this may continue over into the second and possibly even third quarter of 2024. It also explains why you're seeing a lot more reality programming and uh foreign television series on TV and streaming channels, streaming platforms. But there's some other reasons as well. One, as I mentioned, the guilds were prepared. They thought about this deeply in the context of their pre-existing basic agreements and came up with very concrete proposals. Second, their members were already aware of the issues. And as I said, for actors, they knew it was an existential issue. For writers, it was at least a major frontal assault on their livelihoods. The unions worked closely with their members in formulating their proposals. And as a result, morale amongst the striking workers was high from the very beginning. Unlike many unions, the writers and actors guild include a variety of different members. Many of them are paid only at the minimum rate set forth in the contract, known as scale. But they also include A-list players like Tom Hanks and showrunners who both write and produce television series and are very well compensated. Many of these A-list players contributed things like entertainment, food, and beverages to the striking workers while they were on the ticket lines. For example, one week they enjoyed a Taco Tuesday, where someone had contributed a taco truck to feed the strikers tacos. There were also several days when someone provided free coffee and croissants. And there were also theme days, like 80s theme day, where many of the strikers came in costume. There was a lot of music, both live and recorded, as well as the typical sort of speeches that you get at a rally. And you know, when you compare this to the recent United Autoworkers strike, there was a lot more energy behind the writers and actors strikes than you saw amongst the rather the seemingly dispirited auto workers who were walking a picket line at Delantis or General Motors or Ford. Finally, I should note that the studios pretty much lost the public relations battle in this negotiation early on. First, they initially refused to even discuss AI with the writers or actors and said, and publicly took the position that they would not go back to work. I'm sorry, that they would not go back to the bargaining table, that they would not negotiate until the writers had lost their homes and apartments. They, in effect, said they were trying to break the unions. Also, Disney CEO Bob Iger was quoted as saying that the writers were being unrealistic. Might not be so bad if he was being interviewed in his office or over the telephone. But he said this at the Sun Valley Billionaires Conference. The optics were very bad. It was like he told them, let them eat cake. So a variety of things coalesced at the same time to give the unions leverage in negotiating AI guardrails, as well as increases in compensation, residuals, and working time periods, and a streaming video performance bonus.
SPEAKER_00So you just mentioned their compensation. And through this conversation, you've touched on sort of the three C's: consent, credit, and compensation. And then there's the fourth sort of point of disclosure. And you've kind of been touching on those as you've been talking about the threats and the guardrails and the different pressure points that the unions were sort of focusing on with respect to generative AI in particular. I think, you know, if we could just drill down or if we could just sort of at a high level, you know, make sure that people really understand why those were some of the core issues that the studios and the writers and the actors were sort of focusing on through these collective bargaining agreements.
SPEAKER_01Well, let's look to do that. Let's look at the results of the Writers Guild minimum basic agreement. And I'm going to start not with the three C's, but with the big B disclosure. The new WGA basic agreement requires employers to disclose when written material to be used in an assignment or written material added after an assignment was produced by generative AI. Obviously, it's important to know when and what materials are being used because that determines, again, the three C's compensation, the two of the three C's, compensation and credit, but it's more important because only if the use of generative AI is disclosed can the writer reject or give consent to using AI-generated material as part of the writing assignment. Another thing that the basic agreement concluded was that a writer could not be required to revise AI-generated literary material, nor could a writer be required to use AI-generated material as the basis of a screenplay without the writer's consent. So, in short, material generated by AI, generative AI specifically, could not be used without a writer's consent and had to be disclosed in advance so the writers knew what they were consenting to. Conversely, the basic agreement provides that a writer cannot, on his or her own initiative, use generative AI in the course of preparing his or her writing assignment without the consent of the employer. So the studios and production companies, who all seemingly have guidelines about the use of AI, would have to consent before a writer could use generative AI. And obviously, the corollary of this is the writer has to disclose that he or she is using AI in order to obtain this consent. So that's it for the first scene, consent. Let me talk a little bit about the other scenes. As I noted earlier, a lot of the writers' concerns about generative AI were because they had the potential to reduce the writer's compensation, including their reserved rights. In addition, they could also result in a denial or determination of the writer's credit. So those are the second and third fees, compensation and credit. The WGA basic agreement addresses the compensation and consent issues in a few technical ways. And again, as I said, these are based on pre-existing content and concepts from earlier versions of the basic agreement. One, written material produced by generative AI may be assigned to the writer with his or her consent, but it cannot be used and considered as assigned source material for the purposes of determining the level of the writer's compensation for their work. Two, generative AI produced material cannot be considered source material for the purposes of determining a writer's credit and whether the material is original or adapted. That brings us to the third area, which is that generative AI produced material cannot be used as a basis for denying a writer's eligibility for those separated or reserved rights I mentioned earlier. In addition, I should note that the studios in the guild agreed to meet subsequently, at least semi-annually, on the guild's request to discuss the studio's actual and intended use of generative AI in motion picture production and development. So this is part of an ongoing dialogue, but these discussions are not part of the collective bargaining agreement. They can, of course, if the parties agree, result in changes to the collective bargaining agreements, but the studios are under no obligation to accept changes that the writers want to make, and vice versa.
SPEAKER_05Louise, can I make sure that I understand what I thought was a key point to your first example there? Isn't there a term that's used in the collective bargaining agreement that identifies what is original material or there's a special term, and if a writer gets that, the writer gets the compensation, the credit. And if it doesn't, then they start getting kicked down the chain. Am I right about that?
SPEAKER_01Yes. The Writer's Guill Basic Agreement talks about two types of material. One is an original story, which is generated by the writer, either entirely on his or her own, or generated by the writer on the basis of amorphous and very incomplete notes from a production executive. The other is a sign material, which is when the writer is given a pre-existing written work, like a short story, a novel, a screenplay, a biography, a stage play, or something of the like to adapt into a screenplay for film or television. The original story is entitled to separated rights. An adapted screenplay based on assigned material is not.
SPEAKER_05Does that answer your question? Literary material was the term I recall that that's what AI can't get, it can't, whatever it does, it doesn't get to be called that. Only what writers do gets to be called that.
SPEAKER_04So so if I can add a ask another follow-up question, if I think of the the development of a script from some original concept down to something that you can actually shoot a movie from, right? There's this flow that you've described earlier that that starts with the original material and goes through a set of evolutionary stages. And says that AI can't be the source of original material coming in across that bright line and cannot, without agreement between the writer and the and the studio, be involved in any of the steps inside. And no matter what happens, the AI can't subtract the rights of the writers at any stage in the process. Is is that sort of a the way the thing is sort of structured?
SPEAKER_01Yes, that's right. Basically, material contributed by generative AI cannot be considered assigned source material or literary material for determining the level of compensation which the writer gets for his or her work, whether they're paid for a treatment, a first draft screenplay, a rewrite, or a polish. The GI AGI, the GAI generated material is irrelevant. Same thing for credit. Generative AI material cannot be considered source material for determining a writer's credit or eligibility for separated or reserved rights.
SPEAKER_04So, what role, if any, did concerns about copyright have in the strikes and their negotiated settlement?
SPEAKER_01Well, as I mentioned, there are two ways that copyright ownership may flow in a screenplay. One, if it's generated at the instance and request of an employer, the studio, but and under an independent contract, or it's generated by an employee within the scope of his or her employment, it becomes a work made for hire. And that's owned by the party commissioning, the copyright's owned by the party who commissioned that work or who hired the party to write it. In other words, if a writer is hired to write a particular script, the studio or other employer owns the copyright in that script as a work made for hire. Similarly, if a production executive says to a writer, I want you to write a script about a pair of Starquas lovers in divided by controversies and fighting between their families, and outlines a few more of the details, which you might recognize as being the plot for Romeo and Juliet or West Side Story and probably many other stories. But in that case, the writer's guild says you have to hire the writer to do that. But if the writer is not hired under an employment agreement, Agreement, but is hired as an independent contractor, which is allowed under the Writer's Guild Agreement, then that's also considered a work made for hire because audiovisual works are specifically named as one of the types of copyrighted works that can become work made for hire by an independent contractor. There are not many types of works where that's allowed. Motion pictures, other audiovisual works like games are among those. Music is not. So given the fact that either the studios or the writer is the original owner of copyright, the writer would be an original owner if they created a spec script, which then they then assign the rights to the studios. You would think, in theory, that both of them would have the same interest in this, in the matter of authorizing the use of the script to train generative AI and large language models. In other words, they want to make money off of this. They're investing money in it, they're not doing it as a charitable endeavor because they want to contribute to culture, although they do, they're doing it to make money. They monetize the value of all rights under copyright. The problem is that the studios, as we've been talking about them, now include several tech companies like Apple and Amazon. And these companies are investing heavily in their own generative AI for a variety of purposes. Some of those purposes are related to the entertainment industry and reducing cost of production, particularly in the areas of editing, visual effects, dubbing, and the like. Others are not related specifically to the entertainment industry. They include things like customer service, corporate employment and recruiting, preparation of marketing materials. And the studios are not absolutely sure whether or not they want to use AI freely in their creation of written works and the creation of audiovisual works. So most have not taken a stand on whether or not the use of their copyrights to train generative AI constitute infringement. And so far, they generally have not sued. Universal Music, of course, has sued over its music, but it's not affiliated with its motion picture and television divisions. So what the writers wanted was to be able to either authorize the use of their written works as a reserved right to generate additional income for training AI, again, but only with their consent, and only if they get paid what they consider to be a reasonable amount. The studios wanted the right to do this as well, to the exclusion of the writers. So to some extent, you have to determine whether the right to train AI is a reserved right that's included in the contracts or not a reserved right. And who owns it, the studios or the writers. Remember, under reserved rights, separated rights, the writers own the right to publish their own scripts. So they could not really come to an agreement on this point. As a result, they basically agreed to punt. Specifically, they both reserved their rights in this area. The writers would still be allowed to assert claims that the use of their work to train AI is an infringement, but by the same token, so would the studios.
SPEAKER_03If I can jump in here, this has been fascinating and complicated and just generally mind-blowing for a nerd. It doesn't sound I went into this hoping that you were going to tell us all of this and we would come away with the impression that okay, we're done, but it doesn't sound like we're done. They just kick some things down the road.
SPEAKER_05Well, and Louise, isn't it the case that these come up for renegotiation rather soon?
SPEAKER_01As I mentioned, the term of these contracts, the term of the Writers Guild Agreement is about three years. I believe the same is true of the SAG AFTRE agreement. In addition, many of the IATSI agreements expire or have expired early in this year, 2024. So they still need to be negotiated. And there are important issues relating to the use of AI by editors, where some think it's already being widely used, some don't. The use by animators, the use by animation writers who are not covered by the WGA agreement. They're covered by a separate animation guild agreement, which is part of IOTSI. And those are coming up within the, I think they've already started the preliminaries of negotiations for animation. So it will be very interesting to see how people like the Cinematographers Union, the editors union, the animators, and other craft unions like production designers, whether or not they address AI at all, and how they address them. Will they be guided by what the WGA and SAG have agreed in their negotiations? The negotiations for now and the positions of the parties have not been disclosed publicly, so we don't know. And oftentimes, sometimes during the negotiations, those positions are revealed, sometimes they're not, and sometimes the provisions, often the provisions, the actual agreement, the actual terms of the agreement are not disclosed until they're ratified by the members. I should also point out that there are many talented people working in the entertainment industry in fields that are not generally unionized. As I mentioned, there are many productions that are non-union productions, which use non-union writers, actors, and directors. And also productions created by influencers are not subject to union agreements for the most part. So the content produced by influencers for social media platforms like YouTube, TikTok, and Instagram, these guardrails don't apply to them. Also, there are many workers in visual special effects, video games, and the graphic arts for marketing, production design, prop design, costume design, who don't have the benefit of a union. Although many are being unionized, there's a big push to unionize them now. But without a union, they don't have the option of taking their union negotiations and using them to generate guardrails for AI, like the Writers Guild and SAG have done.
SPEAKER_05And I just asked very quickly, Louise, and not to ask you to drill down, but since you're involved in many international film productions and other transactions, could you tell us whether, to your knowledge, any of the other centers of movie making, such as India or Japan or Italy or the UK, are attempting to deal with these issues in such a concerted and focused way?
SPEAKER_01By and large, they have not. First of all, I should note that different jurisdictions have different views of copyright ownership in AI-generated material. So, for example, the owner of AI-generated material can be copyrighted and protected by copyright, at least to some extent in countries like China and South Korea. And also, I believe in the UK, although the extent to which those copyrights cover the material may vary. Second, many of these industries are either completely ununionized or they have unions that are relatively weak in relation to management. So, for example, you don't have a strong Indian actors union or an Indian writers union that can negotiate with the Indian production companies. There are unions in Canada which often piggyback upon agreements made in the US. Sometimes this is done in the UK, but not always. Of course, when an American studio shoots a production in a foreign country utilizing, for example, American actors who are members of SAG, then they are bound by the guardrails in the SAG minimum basic agreement. So I would say it's basically pretty spotty. And as you know, it varies from country to country, the extent to which the government has provided legislation or regulations of AI in general, not just for the entertainment industry.
SPEAKER_00So to wrap us up, and since we have the benefit, Louise, of your experience and expertise in this very complex area, as we've all been discussing, what do you think are the top one or two things that have to happen next to provide appropriate governance for the use of AI in this industry?
SPEAKER_01Well, first of all, as you know from a previous podcast, there are a lot of pending lawsuits claiming copyright infringement or right of publicity infringement in the use of materials to either train AI or create the output through generative AI. We need to get some resolution of these lawsuits, or in lieu of that, legislation that addresses the question is the use of AI to train large language models copyright infringement, if done without the consent of the copyright owner, is output that actually or potentially is substantially similar to the material used for training. Does that constitute an infringement of copyright? We also desperately need a nationwide law to provide for a uniform right of publicity, which I think should include the prohibition of deep fakes in both advertising, entertainment content, and also within the political arena. This should provide for claims for the unauthorized use of name, image, voice, or likeness, not only during a celebrity's lifetime, but also post-mortem. Currently, we have to rely on what's basically a patchwork of state laws, which vary greatly in terms of what they cover and the extent of production and the duration of protection. Not all states have laws on this subject, so it's completely unregulated in some states. But the fact that states like California and New York, which are major centers of production, have these laws, does tend to set in many cases a bar for minimum compliance. California has had a pre- and post-mortem right of publicity by statute for a long time. But New York, for example, has only recently adopted one. So we don't have a lot of case law under that yet. And you know, the other thing that needs to happen, I think, is that attorneys, both in the entertainment industry and outside, need to think about how they're going to draft employment agreements for individual employees to incorporate guardrails for the protection against the risks of generative AI. This is not going to be an easy process as you've seen. It requires consideration of a lot of the details regarding how the work is done and how it's compensated. So I think as attorneys, we have a big task ahead of us in approaching the inevitable rise of generative AI.
SPEAKER_04Wow. Thank you, Louise Nemchoff, for your time and your deep thought and the sharing of your experience and insight. This has been one of the most remarkable conversations, both the preparation that we did several weeks ago and this actual recording session. I came into this process without fully understanding the scope and depth with which the various players in the recent recent sort of combat over AI in the entertainment industry had really sort of prepared for battle and then engaged. And I think the tour you've given us of this issue will certainly have educated me, and I hope it will really provide sort of real depth for our broader audience. And you know, to the objective of the podcast, which is to help educate the broad policy community on how to think about AI and how to think about the governance of AI going forward. I think this will this is a major contribution. Thank you again.
SPEAKER_01Thank you for inviting me.
SPEAKER_00Thank you, we thank the business law section of the American Bar Association for their generous sponsorship of the production of this podcast. We welcome questions and comments from our listeners. Send email to comments at mindthegapdialogues.com. We read all comments and questions and will try to respond in the letters section of a future episode. If you are writing about a particular episode, please do mention the specific episode number. Please do also include pronunciation tips to help us properly say your name when we reply in a subsequent episode. See you next time on Mind the Gap Dialogues on AI.
SPEAKER_02Thank you for listening to the ABA Business Law Section Podcast Series to the extent that the section offers a robust collection of content. To explore more about this topic or to learn about joining the section, visit ambar.org slash bizlaw. That's B-I-Z-L-A-W.