AI and Hollywood: 5 questions for SAG-AFTRA’s chief negotiator

Last summer, Hollywood came to a standstill after the leading entertainment unions went on strike in what amounted to the industry's most significant labour dispute in decades.

One of the unions that decided to strike was SAG-AFTRA, an association that represents the Screen Actors Guild and the American Federation of Television and Radio Artists. The union represents roughly 160,000 actors, broadcast journalists, dancers, DJs, program hosts, singers, stunt performers and voiceover artists, among other media professionals.

SAG-AFTRA chose to strike in July 2023 because of contractual disputes with production companies and streaming platforms over a range of issues such as residuals and the use of new technologies like artificial intelligence (AI). As negotiations advanced, AI became a major sticking point as the union voiced concerns over AI being used to duplicate an actor’s likeness – an ongoing phenomenon that raises a variety of compensation and creative ownership dilemmas.

“We came in saying we're willing to partner with you on AI, but there have to be guardrails and protections built into the contract,” Duncan Crabtree-Ireland, SAG-AFTRA’s National Executive Director and Chief Negotiator, told the World Economic Forum last month during its Annual Meeting 2024 in Davos, Switzerland.

Utimilatey, the strike ended in November 2023 after SAG-AFTRA and the studios reached an agreement. Yet AI and entertainment industry experts note that the issue of AI in entertainment is likely to reemerge as the technology advances.

In the following question-and-answer, Crabtree-Ireland further details the influence of AI in the strike and the technology’s wider impact on Hollywood.

“We knew it was going to be super important for our members. Every year we go to a trade show in the US called the Consumer Electronics Show. This is a place that we have over the last two decades used as a way to see a little bit around the curve and know what's coming. So because of our involvement there, we really started focusing on AI and the potential of generative AI a number of years ago.

“So when we started this negotiation in June of last year, we knew that it was going to be a big issue. We had an extensive proposal on AI, but what we didn't know was that the companies would be so insistent on refusing to agree to what we thought were really reasonable proposals. We came into this not looking to ban AI. We didn't take that approach because we feel that approach is not likely to be successful. Past history tells us that.

“So we came in saying we're willing to partner with you on AI, but there have to be guardrails and protections built into the contract, and it ended up being very difficult to get the studios or streamers to agree to the kinds of protections we needed. And that is something that ultimately only happened several months into the strike, when the CEOs of the studios and streamers got directly involved, and we really started hammering out what would become the ultimate deal on AI.”

“For our members, who are primarily actors, the concern was there's already AI being used to digitally replicate actors. It's something people may not realise, but they may be familiar with when Carrie Fisher was replicated for Star Wars or when Paul Walker was replicated to finish out the Fast and Furious movie that he was in the year that he passed. These are things that were done with early iterations of AI.

“So it was clear to us that as these tools become more capable and less expensive to use, their use was going to grow – especially because what these tools do is they take someone's image, their likeness, their voice, their performance, and they turn it into something that they never participated in the creation of. That's a very personal thing. It's not only about having a job, it's also about the fact that someone is using your face and your voice to deliver a performance or a message that you had nothing to do with.

“So we knew we had to have guardrails in place to prevent that. And ultimately we succeeded in achieving those. So in this contract, there are provisions that say the studios and streamers can't do anything using AI or digital replications without the informed consent of the performer. So you can't put words in someone's mouth without their approval. You can't use them for things that they haven't signed off on, and it can't be some generic line in a contract that says, I hereby give you permission to just do whatever you want with my image or likeness. It has to be specific, detailed information about the intended use.

“Another area of concern that will likely come in the future is the creation of fully synthetic performers. It's one thing to scan a performer – maybe train an AI using that particular performer’s past performances and then create a performance that reflects them – it's another thing to take a generative AI system and train it with thousands or tens of thousands of performers, and then have it create a new performer who doesn't have a corresponding human being. What does that mean? How does that affect jobs? How does that affect fair compensation? And is it really fair to even call that a performer when what it actually is an AI-driven synthesis based on a bunch of other performers' creative work?”

“Absolutely – and this is an ongoing process. I think, because when people think about AI, they have a certain image in their mind. For a lot of people, it's Terminators or Skynet. And even for professionals working in this industry, they haven't had the opportunity to educate themselves about what these tools do, how generative AI in particular might be used. So we had to make sure that our members had the chance to inform themselves so they can make solid decisions about how to address AI.

“In the case of this negotiation, our negotiating committee unanimously endorsed what was ultimately negotiated in this process. And then we had a very robust discussion with our members. It lasted over three weeks. At the end of that, about 80% of our members voted in favour of this agreement.

“Still, I think it's important to recognize that there's a sizable minority who do have issues with it. In that 20%, some of those people really want us to just try to block AI. I understand the sentiment, but I also believe firmly that strategically, that's the wrong approach. Trying to block the adoption of this technology will essentially mean that we walk away from our opportunity to channel its direction. Blocking technology has never proved to be effective and I don't think it'll prove to be effective here.”

“Absolutely. I think it's clear that the capability of AI is going to evolve. The desire of companies and how they want to use AI is also going to evolve, and our contractual protections and provisions are going to evolve.

“But I'm confident that the protections we have in place now will serve us well during the course of this contract. We will be renegotiating this deal in less than two and a half years. At that time, we'll have the benefit of seeing what's happened over that time period. Part of this agreement is that we are entitled to have meetings every six months with all of the companies to find out what they're doing in AI, and specifically with generative AI. So having had those meetings by that time, I think we'll be in a strong position to make sure that we're channelling our efforts in the right direction.

“But I think ultimately, if we put all of our power and force into trying to block AI, which would ultimately be an unsuccessful effort, we then will have given up the chance to really nudge it in the right direction. And I do think putting guardrails up and pushing AI into the right kind of implementation is how we can ultimately see a human-centred use of AI in the entertainment industry instead of something that's dehumanising or devaluing creative talent.”

“I think first of all, educate yourself. Before you launch into a negotiation addressing AI issues, you've got to go out and use whatever tools and resources you have, it could be the AFL-CIO, other unions like us, your own investigation or experts that you hire. You have to build the knowledge and expertise that you need in order to be on a level playing field with these companies.

“Second, you have to take these issues head-on because the fact is, whatever you don't negotiate as part of your contract, the companies will attempt to get around it in any industry. You need to have a direct conversation with the companies about what is and what is not okay with the workers.

“And number three is really for all of us, which is we need to set a standard that workers are included in all conversations around AI’s implementation. When people talk about public-private partnership, they often think about the corporate world and then government. Who's missing in that equation is workers. So we need to set a standard for workers’ inclusion.

“In the case of our strike, if there had been real dialogue and collaboration between the studios, streamers and unions, we might have been able to avoid those strikes. Instead, we had strikes that cost the entertainment industry in the US more than $6.5 billion. Hopefully others will learn from that lesson and realize that cooperation and collaboration is needed – and that AI implementation must be human-centred in every industry.”