"AI in the Arts Is the Destruction of the Film Industry. We Can't Go Quietly"

by Justine Bateman 

[Justine Bateman’s op-ed appeared on the Newsweek website on 17 May 2023 (AI in the Arts Is the Destruction of the Film Industry. We Can’t Go Quietly | Opinion (newsweek.com)).  As a working member of the entertainment industry and someone with both academic and practical knowledge of artificial intelligence and computer-created output, she clearly has strong feelings about the practice and takes a stand.]

What does it mean to be human?

You look human, you act human, you learn lessons, you have challenges, you feel emotions.

And yet, in 2023, we’ve shrunk decidedly away from being human.

The Writer’s [sic] Guild of America (WGA) is currently on strike against the AMPTP, the representation of the Hollywood studios and streamers. A number of demands were made and were met with the expected pushback, but one pushback was alarming: the refusal to even have a conversation about the potential for AI to displace screenwriters in films and series.

As a WGA writer, a Directors Guild of America (DGA) director, a former Screen Actors’ Guild (SAG) board member, former SAG negotiating committee member, and coder who holds a UCLA degree in computer science and digital media management, I knew this signaled that they were not only thinking about using AI to displace us, but that they had already begun.

AI stands for Artificial Intelligence, but I refer to it as “Automatic Imitation.” In short, AI is an algorithm that is fed a wealth of information and given a task, and it then delivers the result based on the information it’s been fed. There are more complexities, but that is the basic design and function of AI. And it is being used in the Arts for greed, trained on all our past work.

This is how I believe it’s going to play out:

It starts with AI-written scrips and digitally-scanned actors, either image or voice actors. This scanning is already in practice; in fact, some talent agencies are actively recruiting their clients to be scanned. What this would mean for the actor is that they would get 75 cents on the dollar, and their digital image can be triple and quadruple booked. Of course, you’re not getting the actor; you’re getting a copy of them.

The next step will be films customized for a viewer based on their viewing history, which has been collected for many years. Actors will have the option to have their image “bought out” to be used in anything at all. Viewers will be able to “order up” films—for example, “I want a film about a panda and a unicorn who save the world in a rocket ship. And put Bill Murray in it.”

From there I believe viewers will be given the ability to be digitally scanned themselves, and pay extra to have themselves inserted in these custom films. You’ll also start to see licensing deals made with studios, so that viewers can order up older films like “Star Wars” and put their face on Luke Skywalker’s body, and their ex-wife’s face on Darth Vader’s body, and so on.

You can also expect to see the training of AI programs on older, hit TV series in order to create new seasons. “Family Ties,” for example, has 167 episodes, comprising seven seasons. An AI program could easily be trained on this to create an eighth season.

All to say, AI has to be addressed now or never.

I believe this is the last time any labor action will be effective in our business. If we don’t make strong rules now, they simply won’t notice if we strike in three years, because at that point, they won’t need us.

The future I’m describing rings true for many, though some have told me that they don’t believe that viewers want to see AI-generated images, or see themselves in AI films, or watch regurgitations of past films.

I believe they are wrong: Viewers have already been conditioned for AI film, because we have gotten away from being fully human.

If we were in 1975 and we asked people what they thought of these AI images, they’d most likely furrow their brow and remark on its artificial nature. But in 2023, our eyes have been trained on faces that have been amended by plastic surgery, by Instagram and TikTok filters. People are now more accustomed to seeing artificial faces than they are fully natural human faces.

Manufactured AI images are just one click away on the false imagery dial.

As far as “seeing themselves in AI films,” I believe society is well-conditioned for that as well. For the past 10 years, social media apps like Instagram and TikTok have opened wide the narcissism and self-obsession usually dormant in humans. The vast majority of what’s posted on social media is an almost frantic stream of selfies and vlogs, all posted in the name of “building your brand.” Seeing yourself in custom made films would easily fit into this model.

But we’ve also been primed in another way for the AI takeover of art: Endless reboots, remakes, sequels, next chapters, and prequels have replaced new stories. We’ve basically be doing AI by hand, pulling in all the old stuff and spitting out an amalgamation of that past work.

The creation of new paths and genres, so important to being a modern human, have not really occurred in the Arts for the past 20 years. What do you think of when you think of 21st century art? Most of it is just a rehashing of the 20th century. And this makes people open to the rehashing of the past that AI is designed to do.

As much as I dislike it, I believe society in 2023 (and 2024 and 2025) will be receptive to AI films and will embrace their arrival. And I believe the studios and the streamers know this. I hope I am wrong about this, but I don’t think I am. And it’s unfortunate. AI in the Arts is nothing less than a destruction of the 100-year[-]old film industry.

We must fight, we must strike as WGA members, we must warn other artists—actors, crew members, and directors—about what is coming. Because this is our watch.

All of the film industry that has come before us handed us this business, and it was ours to care for. If the studios and streamers that hold the purse strings that enable our work are set on destroying this incredible business that brings so much to society, I want it known that on our watch, we opposed it. We defended human-made films and series.

On our watch, we fought being replaced with computer programs.

We must fight, we must strike as WGA members, we must warn other artists—actors, crew members, and directors—about what is coming. Because this is our watch.

[Justine Bateman is a filmmaker, author, and coder with a degree from UCLA in computer science and digital media management.  In addition, she’s a film actress, writer, director, and producer.  (She’s also the older sister of actor, director, and producer Jason Bateman.)

[The views expressed in this article are the writer’s own.]

*  *  *  *

[Bateman, as an artist as well as someone knowledgeable about computers and AI, is writing from her individual perspective.  She also focuses on the consequences, as she sees them, of the misuse and abuse of AI technology in film and television production. 

[In the Spring 2023 issue of SAG-AFTRA magazine (vol 12, no. 2), the film-and-TV actors’ union lays out the issues facing performers in the coming technological world of AI in an article entitled “Artificial Intelligence, Real Consequences.” 

[“Of all the tasks that could be automated,” admonishes the article, “generating art was perhaps the one that was least in need of being done by a machine, but now that the technology has emerged, it’s going to be part of the new reality for all creators.”  Asserts one union official: “Creative storytelling is fundamentally human, and humans should remain at the center of the story.”

[SAG-AFTRA is aimed at an audience of union members, so I thought that I’d present portions of it that are of more interest to a general readership on this blog (leaving out some of the more inside-baseball reportage—with apologies to the SAG-AFTRA leadership):

It’s a concerning time for creatives. Generative artificial intelligence has burst onto the scene, allowing anyone to create essays, screenplays, pictures, music and more in mere seconds, with minimal effort — even if they consider themselves devoid of creativity.

The most popular of these tools parse a user’s input and, using an enormous set of data, create an output that fulfills that request. So, for instance, ChatGPT can respond to a user’s request to “Write an essay about snails” by pulling everything it knows about the invertebrates to create a passable article on the topic. On the visual arts side, one might ask Midjourney or DALL-E for a picture of a snail in the style of Vincent Van Gogh, and the AI will dutifully spit one out.

Machine learning, which powers these tools, is being used onscreen to de-age actors, lip-synch dubbed programs, replicate an actor’s voice and even generate entire performances. It’s not hard to see how all this power can be misused.

That’s why SAG-AFTRA and other industry unions are working hard to ensure that AI is available as a tool to help creators, not a way to put them out of work.

“These tools can help America’s entertainers ideate and work more efficiently. They can be a welcome tool to assist in the creative process, or they can take a darker turn, threatening to replace the tens of thousands of hardworking creative minds that have dedicated their lives, and their careers, to their crafts,” said SAG-AFTRA NED Duncan Crabtree-Ireland. “In my time as SAG-AFTRA national executive director, I’ve seen new technologies contribute to, rather than disrupt, the work that our members do for the public. With the right approach to regulation, we can ensure that new advancements in generative AI do the same.”

To that end, the union has staked out a firm position on the use of AI and is taking concrete steps to ensure the technology is used for good. Without the power of collective action, performers would be on their own in navigating this new threat, and would have to rely solely on the benevolence of those who hire them.

. . . .

“Creative storytelling is fundamentally human, and humans should remain at the center of the story. SAG-AFTRA has always embraced technological change, but we have done so with our longstanding mandate to protect actors, recording artists and broadcasters foremost in mind. We will work with our industry partners to negotiate fair terms and conditions for AI uses, and we will partner with them to advocate for smart AI policies. Most importantly, we will never stand idly by if the professional voices, likenesses and performances of our members are misappropriated,” said SAG-AFTRA General Counsel Jeffrey Bennett.

In March, the union announced that it joined with major industry players to support the Human Artistry Campaign [see below] and that the rights to digitally simulate a performer to create new performances must be bargained with the union. SAG-AFTRA warned that any attempt by employers to circumvent the union and deal directly with the performers on these issues is a violation of the National Labor Relations Act. SAG-AFTRA further clarified that Global Rule One, which requires that SAG-AFTRA members must always work under a union contract wherever they work, covers any agreement with an employer to digitally simulate a member’s voice or likeness to create a new performance.

As the technology improves at a dizzying speed, the laws protecting copyright and other intellectual rights will be tested. SAG-AFTRA emphasizes that governments should not create new copyright or other IP exemptions that allow AI developers to exploit creative works, or professional voices and likenesses, without permission or compensation. AI that generates text and art doesn’t create it from nothing; it is trained on the hard work, brilliance, inspiration, sweat and creativity of countless artists — artists who have financial obligations and families, and who deserve to be compensated for their efforts.

[“The act of creating is one of the most fulfilling and satisfying aspects of human existence,” declares the article, “and SAG-AFTRA is doing everything it can to keep it that way.”  Toward that end, as the excerpt posted above reports, the union has joined with other performing-arts organizations “to ensure AI is used responsibly.”  The Human Artistry Campaign’s fundamental principles are:

1. Technology has long empowered human expression, and AI will be no different.

2. Human-created works will continue to play an essential role in our lives.

3. Use of copyrighted works, and the use of voices and likenesses of professional performers, requires authorization, licensing and compliance with all relevant laws.

4. Governments should not create new copyright or other IP exemptions that allow AI developers to exploit creators without permission or compensation.

5. Copyright should only protect the unique value of human intellectual creativity.

6. Trustworthiness and transparency are essential to the success of AI and protection of creators.

7. Creators’ interests must be represented in policymaking.

[To take action, visit https://www.humanartistrycampaign.com/.]

"Use of artificial intelligence generates questions about the future of art"

by Jeffrey Brown and Alison Thoet 

[The PBS NewsHour aired a feature on the broadcast of Thursday, 2 March, that reported on the latest incarnation of artificial intelligence.  Last Sunday, 5 March, Lesley Stahl presented a report on chatbots, AI programs that can write a human-sounding essay in less time that you and I can write a sentence.  All last month, there were reports on news programs and in newspapers covering the release of ChatGPT, a new, powerful chatbot.

[There are serious problems with the writing AI, more than the obvious issue of students and reporters and others who must write using the programs to do their work for them, and perhaps getting away with it.  The chatbots can get facts wrong—and, in effect, they can lie.

[The NewsHour report, however, is about AI in a different field: art.  We’ve seen stories of “art” purportedly created by infants or chimpanzees or even elephants, but those were usually gimmicks, even if they weren’t hoaxes.  Now we may see art become a field in which human artists have to compete with computers.

[Is it art if it’s created by a machine?  We’ve had computer art for a long time already, but that’s been artists using the computer as a tool in a sense, like a brush and palette or a camera.  The machine didn’t create, not on its own.  In the current instances, the artist is still the moving force in the art’s creation, but it seems to be a step closer to machine-generated work.  Is it a threat to artists or the art world?  We’ll have to see, I guess.

[I’ve posted a number of articles on Rick On Theater on the subject of computer technology, including AI, inching its way into the art of theater.  Some of it’s already obsolete—and the blog’s only going on 14 year old (on the 16th, as it happens).  Have a look at “Theater and Computers,” 5 December 2009; “‘A Guinea Pig’s Night at the Theater’” by Dave Itzkoff, 28 January 2013; “Technologies Old and New,” 9 March 2018; “Computers and Actors,” 4 and 7 October 2021; “‘Entertainment in the Age of AI,’” 22 August 2022.]

Artificial intelligence is everywhere and part of our conversations about education, politics and social media. It’s also a hot topic in the arts world as programs that generate art using AI are widely available to the public. But what goes into these programs, and the work that comes out, are heavily debated. Jeffrey Brown explores the influence of AI for our arts and culture series, CANVAS.

Amna Nawaz: Artificial intelligence, or A.I., is everywhere. It’s now part of our conversations about education and politics and social media. It’s also become a hot topic in the art world.

Programs that generate art using A.I. are widely available to the public and are skyrocketing in popularity. But what goes into these programs and the work that comes out are heavily debated in the arts community.

Jeffrey Brown explores the influence of A.I. on art, and where it may be headed next. It’s part of our arts and culture series, Canvas.

Jeffrey Brown: An enormous 40-by-14-foot screen, constantly morphing abstraction, created from some 35 million images of coral from around the world.

What are we looking at?

Refik Anadol, Artist: These are machines’ interpretation of millions of data of underwater.

Jeffrey Brown: It’s a kind of artificial reality, in the term of artist Refik Anadol, who calls this a data painting.

Refik Anadol: It’s in flux, doesn’t dry. The pigment is always in movement. The colors are changing.

Jeffrey Brown: The 37-year-old Turkish born artist in perpetual motion and overflowing with energy at the opening of a new exhibition of his work at the Jeffrey Deitch art gallery in Los Angeles, is at the leading edge of a growing artificial intelligence, or A.I., art revolution.

Refik Anadol: My primary material is data, the information, and my second material is a thinking brush that is assisted . . .

Jeffrey Brown: Thinking — a thinking brush?

Refik Anadol: Yes, that is assisted by artificial intelligence.

Jeffrey Brown: Anadol’s artworks are made in a Los Angeles studio in which computers, software programs, screens and a 3-D printer with a giant robotic arm replace more traditional artist’s materials.

Here, he and a team, including computer and data scientists, engineers, researchers, architects and designers, build enormous data sets requiring huge computing power — you can’t do this at home — and run them through or train A.I. programs they themselves have created.

The input can vary. Right now, at New York’s Museum of Modern Art, Anadol has created a piece titled Unsupervised created from images, research and other data from the museum’s entire archives. In another example back at his studio, the input is a whole lot of flowers.

This is your brush in this case.

Refik Anadol: Yes. It’s like with a game joystick.

So, what we are watching here is an A.I. that is trained on 75 million flowers.

Jeffrey Brown: Seventy-five million images of flowers.

Refik Anadol: Of flowers, exactly.

Jeffrey Brown: And then the A.I. . . .

(CROSSTALK)

Jeffrey Brown: . . . is making its own flowers.

Refik Anadol: Yes, exactly.

And — but to make it happen, it’s when and machine and human connection happens.

Jeffrey Brown: We’re not seeing real flowers here, but an artistic collaboration of machine and human.

Who is the artist?

Refik Anadol: Artist is still the artist. Like, human is the human there.

Jeffrey Brown: You are the artist?

Refik Anadol: Yes. But let’s also remember that it’s an — assisted by a machine, an A.I., an algorithm. And I think this is really inspiring to me, because that’s exactly where art, science and technology, a beautiful movement that I hope it’s inspiring. And it’s most likely our future, where we are more getting closer to machines every single day.

Jeffrey Brown: You refer constantly to the machine dreaming.

Refik Anadol: Yes.

Jeffrey Brown: Is the machine thinking too?

Refik Anadol: No, I don’t think machine is thinking. Machine is helping to think in a different way.

Jeffrey Brown: It’s the stuff of sci-fi, yes, but art and technology have been connected and debated forever, including in the early decades of photography.

Dhyandra Lawson is an assistant curator of contemporary art at the Los Angeles County Museum of Art, known as LACMA.

Dhyandra Lawson, Los Angeles County Museum of Art: Photographs were dismissed and marginalized from fine art discourse. It’s because of the presence of the apparatus, the machine, the camera, right, which people had claimed, critics claimed, took away the artist’s special hand.

Jeffrey Brown: And the computer itself has been a fascination for artists for more than 50 years, made clear in a LACMA exhibition titled Coded: Art Enters the Computer Age, artists playing with the physical machine and exploring its ability, primitive compared to today, to generate images.

Coded brings the story up to 1982, the advent of the Internet age.

Dhyandra Lawson: What is new now I think is access.

Jeffrey Brown: Access, that is, for everyone, including to a new kind of artmaking on A.I.-generating apps available to the wider public, offered by companies such as Midjourney, DALL-E, Stable Diffusion, and a proliferation of others, which allow the artist in each of us to create an artwork through simple prompts.

It’s easy and fun, but has also raised a host of new questions. Last year, for example, an A.I.-generated work won a prize at the Colorado State Fair, outraging other contestants. And some artists, like Molly Crabapple, see another, more urgent threat.

Molly Crabapple, Artist and Author: I see a world where illustrators like me, people who make our living drawing pictures, are completely replaced by these apps because they can make images faster and cheaper than any human possibly can.

And, even worse than that, this generative A.I. was trained on our stolen images.

Jeffrey Brown: Crabapple is a New York-based artist and writer. Her critique? The new platforms build their businesses by scraping, or collecting, images and other data from the Internet, including works by artists and illustrators who aren’t credited or compensated.

And that, she says, is already changing the landscape for her community.

Molly Crabapple: There are companies that, even one year ago, were hiring artists to do book covers and creating beautiful book covers with those artists that are now making A.I. book covers.

Jeffrey Brown: In January, in fact, a group of artists filed a class action lawsuit now pending against several A.I. imagery generators, charging copyright violations.

I asked Crabapple, who’s not part of the suit, about A.I.’s lasting impact on the art world.

Molly Crabapple: Do I think that human creativity will be destroyed because there are these little image generators? No. People are creative.

However, do I think the lives of individual artists who are making their income will be destroyed if they no longer have work? Yes, of course I do.

Jeffrey Brown: Another concern in these early days of A.I. art suddenly being available to all, the images themselves, and how much they reflect the biases, or worse, of the Internet databases from where they have been obtained.

Curator Dhyandra Lawson counsels that consumers need to consider the source of the data being used to generate A.I. art.

Dhyandra Lawson: The things that we need to be aware of are, who and what is training the A.I.? What sort of data are they — is it using? Does the data perpetuate power dynamics and inequalities in our — in our lived experiences, in our real lives?

Jeffrey Brown: Artist Refik Anadol agrees with such concerns. His work, he points out, differs from widely available apps, in that he and his team generate their own data sets and take care with their sources, what is sometimes referred to as ethically sourced data.

He remains a big believer in the future of this technology. He’s convinced that the work going into his art, for example, the mapping of coral reefs, can help with environmental and a host of other big problems in the world.

But, he cautions:

Refik Anadol: We should think A.I. as a mirror. A.I. is a mirror, and this mirror can reflect whatever we are training. So, if we are aware of this mirror that reflects the information, I think we have more clarity.

It’s really on our hands to train this mind to dream what we want. It’s — the human intention is here. It’s not the A.I., I think, the issue.

Jeffrey Brown: The human, that is, is, for now, still the more complicated machine.

For the “PBS NewsHour,” I’m Jeffrey Brown in Los Angeles.

[In his more than 30-year career with the PBS NewsHour, Jeffrey Brown has served as co-anchor, studio moderator, and field reporter on a wide range of national and international issues, with work taking him around the country and to many parts of the globe. As arts correspondent he has profiled many of the world’s leading writers, musicians, actors, and other artists. 

[Among his signature works at the NewsHour: a multi-year series, “Culture at Risk,” about threatened cultural heritage in the United States and abroad; the creation of the NewsHour’s online “Art Beat”; and hosting the monthly book club, “Now Read This,” a collaboration with The New York Times.

[Alison Thoet is a writer and a CANVAS associate producer and national affairs associate producer at NewsHour.]

"Entertainment in the Age of AI"

 

[The following article and two sidebars appeared in the Summer 2022 issue of SAG-AFTRA (11.3), the quarterly magazine of the union that represents actors in television and film. 

[Readers should note that I have published past posts on the subject of the use of computer and advanced technologies in TV, film, and theater.  I refer to “Theater and Computers,” posted on Rick On Theater on 5 December 2009, and “Computers and Actors,” 4 and 7 October 2021.

[For those who aren’t part of the performance industry in the United States, SAG-AFTRA, the result of the merger of two formerly independent unions, stands for Screen Actors Guild-American Federation of Television and Radio Artists.]

Imagine waking up to find you are the face of a new advertising campaign — and it’s a product you don’t want to be associated with. That was played for laughs on an episode of Friends, when Joey did some modeling and then, unbeknownst to him, ended up the face of an STD campaign [“The One Where Underdog Gets Away”; Season 1, Episode 9; 17 November 1994].

As technology has evolved, artificial intelligence-powered software has made it possible to create realistic audiovisual, video and audio content known as “deepfakes.” It makes the above scenario not only possible, but a real threat to those who sign broadly written non-union contracts that allow for unfettered use of a performer’s image or voice.

In 2018, SAG-AFTRA magazine reported on the growth of digital replicas’ threats and opportunities. At that time, it was still a relatively niche technology, but in the last few years it has gone mainstream. Nonconsensual deepfakes remain a problem — one the union remains vigilant about — but the underlying technology has many legitimate uses that can provide exciting new opportunities for members.

AI-Generated Content Is Growing

In recent years, there has been an explosion in the number of artificial intelligence, or “AI,” content technologies, and the quality of AI-generated content has improved exponentially.

AI tech has been used in large- and small-budget entertainment projects to virtually age and de-age characters in a way that is cleaner, cheaper and more believable than traditional visual effects, and without countless hours in the makeup chair. AI can simplify performance capture, potentially eliminating the need for capture suits and head rigs altogether. It can even be used to enhance the work performed by stunt performers, which can provide safety benefits.

In dubbing and ADR [presumably automated dialogue replacement, a post-production process in filmmaking], AI technologies can help match an actor’s mouth and facial movements to the dialogue they are speaking. It can also be used to dub the films themselves. In the context of projects originally produced under SAG-AFTRA agreements, it could open new revenue opportunities by providing members an opportunity to negotiate for their voices to be used in the foreign-language release. On the other hand, distributors of foreign content who would otherwise hire SAG-AFTRA members to do dubbing work might find it preferable to use AI-generated audio of the original actors. Although this has not yet happened on a widespread scale, there are companies proposing these business models as the technology improves, so the union is monitoring this closely.

Outside entertainment, AI-generated audio can be used in digital assistant devices, customer service, speech assistance and countless other applications, opening new areas of work for professional voice performers.

AI technologies have brought historic figures back to “life” in education and museum settings, typically with the help of an actor who provides the performance capture that animates the digital person.

This field is full of exciting innovations, and a lot of performers are eager to work in these new areas and potentially generate new income through their AI voice or avatar, but it’s important to understand both the technology and the pros and cons of working with it.

“Technological innovations have historically provided incredible new opportunities for our members,” said SAG-AFTRA Executive Vice President Ben Whitehair. “But we must, as we always have, be deeply mindful of the associated risks, and ensure that our digital performances and likeness are protected.”

Know What You Are Agreeing To

Anytime you grant rights to use your name, image, likeness and voice, you should have a clear contract in place governing the use. This is even more important in the digital context. But, even on traditional entertainment projects, performers are often asked to grant rights to use their voice, likeness, and performance well beyond what is necessary for the specific project.

You might have heard about a lawsuit filed by a Canadian voice actor against the company behind TikTok. The performer had done voice work for a Scotland-based company, but the voice files were allegedly used without her consent in the popular app. The case illustrates the risks for actors, particularly when working without the protection of the union behind you. Being branded as an app’s voice and being involuntarily associated with content that you cannot control can impact your image and ability to attract other voiceover work.

The nonunion AI contracts SAG-AFTRA has seen have very one-sided terms and are often with companies based in foreign countries. Many of these contracts give broad rights to use your likeness or voice irrevocably and in perpetuity — this means they have those rights forever and you cannot cancel the permission. There typically is no case-by-case approval over how your digital self is used, and no form of residuals or use-based payments no matter how long or widespread the use is.

“There are contract templates floating around that AI companies and industry players claim are performer-friendly, but a close look at the terms reveals a lack of crucial protections,” said Senior Assistant General Counsel, Compliance and Contracts Danielle Van Lier [see a related article by Van Lier below].

Rights of publicity — the laws that protect your name, voice, image and likeness — can potentially help against unauthorized uses of your digital self. These laws and others do not provide a remedy when you sign a contract granting away the rights. Without a union contract covering your work, your only potential recourse is costly litigation.

The entertainment and media industry is always evolving, and as technology advances, it seems that the pace of change is ever-increasing. SAG-AFTRA members and staff are continually working with tech firms, attending conferences and staying up to date on all the latest information in this emerging field to ensure members are protected.

“Artificial intelligence is opening new frontiers in digital manipulation, and while it is new territory, it doesn’t have to be scary, as long as we stay informed about the potential hazards,” said SAG-AFTRA President Fran Drescher. “As we move into this bold new future together, your union will be standing by your side.”

*  *  *  *

SAG-AFTRA and AI

SAG-AFTRA has been working on issues relating to digital avatars and voices since long before AI was being used to create them.

THE UNION has several staff members with considerable experience and expertise on name, image and likeness rights, as well as on AI technology and its applications, deepfakes and other related topics. They have presented to a global audience on these subjects, hosted panels and discussions with experts in the field, and have written and been interviewed for numerous articles related to AI. SAG-AFTRA also participates in a multi-union workgroup on AI with British Equity and ACTRA [Alliance of Canadian Cinema, Television and Radio Artists]. 

Anticipating the rise of digital replicas, SAG-AFTRA has added or negotiated language into many of its promulgated and collectively bargained agreements prohibiting the creation or use of digital replicas without both the union’s and the performer’s consent, including those covering audiobooks, video games, podcasts, commercials and corporate/educational content.

“SAG-AFTRA HAS been advocating for rights of publicity and name, image and likeness rights for decades. This includes supporting critical legislation as well as writing amicus briefs in cases that could impact how laws relating to these rights are interpreted,” said SAG-AFTRA General Counsel Jeff Bennett. SAG-AFTRA drafted and was instrumental in California’s and New York’s civil laws against unauthorized digital nudity, as well as the passage of New York’s new right of publicity law, which includes prohibitions on using digital avatars of deceased performers.

FOR SEVERAL years, SAG-AFTRA has been in conversation with AI technology companies about the ethical use of this technology and the fair compensation and protection of performers who allow their voice, image or performance to be used in the development and use of an AI voice or avatar. 

“SAG-AFTRA is committed to ensuring that our members’ rights are safeguarded and that they are paid what they deserve whenever their work is exploited, regardless of the technology employed or the nature of the exhibition platform,” said Senior Director, Strategic Initiatives Sue-Anne Morrow.

All SAG-AFTRA contracts for work in the AI space include these critical terms:

•  Safe storage of the performer’s voice, likeness and performance, and the products and content created from them.

•  The right to consent — or not consent — to uses.

•  Explicit limitation on use of the content.

•  Appropriate payment for use of the content.

•  Any exclusivity must be clearly noted and fairly compensated.

•  The right for a performer to control or opt out of continued use and production.

*  *  *  *

What You Can Do

SAG-AFTRA is actively engaging in discussions with companies creating AI content, and has crafted contracts that are relevant to this new work area, are easy to use and provide protections for both the performer and the employer. But your help is critical in establishing a strong foothold in these emerging spaces.

Don’t work off the card: Many AI companies have websites through which anyone can submit their voice or likeness and sign up to be an AI “spokesperson.” Working in this space without a SAG-AFTRA contract is not only a dangerous move for a professional performer, it also impedes the union’s efforts to set fair terms and protections. [“Off the card” refers to the union membership card.]

Let your union help: If you are approached to do this kind of work, ask your employer to consider hiring you under a union contract. SAG-AFTRA staff is happy to talk to them and make the process of becoming a signatory simple and easy.

Talk to your peers and students: Let your peers know that their best protection, when working with AI technology, is a union contract. If you teach classes to or mentor actors who are early in their careers, warn them of the risks discussed in this article. 

Communicate with your representatives: Ask your professional representatives if they are current on the technology and understand the risks associated with it, and let them know that you aren’t interested in venturing into this space without your union behind you.

*  *  *  *

[The article above from SAG-AFTRA magazine included a note at the end that SAG-AFTRA Senior Assistant General Counsel, Contracts and Compliance Danielle Van Lier provided guidance for lawyers who are representing performers in connection with AI-generated content in the May 2022 issue of Los Angeles Lawyer magazine.  I’ve decided to append that article to this Rick On Theater post.]

Practice Tips:
PROTECTING ARTISTS’ RIGHTS IN THE AGE OF AI
by Danielle S. Van Lier

[Van Lier’s article appeared in the May 2022 issue of Los Angeles Lawyer (45.3), the magazine of the Los Angeles County Bar Association.]

Two years ago, in this magazine, another article was published about the rise of deepfakes and their potential for abuse. At that time, California had recently enacted Civil Code Section 1708.86, which created a civil cause of action for individuals who, without consent, are digitally depicted as “giving a performance they did not actually perform” in “any portion of an audiovisual work that shows the depicted individual performing in the nude [as defined] or appearing to engage in, or being subjected to, sexual conduct.” The article discussed the potential harm to, among others, performers when they are involuntarily depicted in the nude or as engaging in sexual conduct without their consent. A lot has changed in that short time.

“Deepfake,” a portmanteau [a blend of words in which parts of multiple words are combined into a new word] of “deep learning” and “fake,” has become the most prevalent term used to describe audiovisual and audio content created using artificial intelligence (AI). These videos typically depict someone doing something they did not do, or saying something they did not say. The term originated in connection with nonconsensual pornography and, from its onset, some have questioned whether the term is overused and its definition too amorphous. For purposes of this article, use of the term “deepfake” is limited to nonconsensual content or that created with the intent to deceive, while using “AI-generated” to refer to content created consensually.

Deepfakes remain a problem and a threat. No matter how seemingly benign, nonconsensual deepfakes can harm the individual whose voice and/or likeness is used. These concerns go beyond nonconsensual sexual content to uses in a commercial or even creative setting. It can harm the depicted individual’s reputation, mislead viewers and consumers, or foreclose job opportunities.

For all the potential harm deepfakes cause, the underlying technology has many legitimate use cases. However, even authorized use can go too far if one is not careful in the contract process. SAG-AFTRA has been closely watching the development of these new technologies and the agreements being used in the space. A number of issues have surfaced under this review, which have resulted in the formulation of some questions to ask if a client is approached to work on an AI-project.

Growth of AI-generated Content

There has been an explosion in the number of AI technologies, like the ones used to create deepfakes, and the quality of AI-generated content has improved exponentially. Companies offering AI-generated people and audio have proliferated in recent years.

The technology has potential for positive applications. In the entertainment industry, for example, it can give independent producers with lower budgets some of the same capabilities as the major motion picture studios. It can help match an actor’s mouth and facial movements to dialogue in foreign-dubbed films, or even to dub the films themselves. In a recent episode of The Book of Boba Fett [streaming Star Wars series on Disney+, 2021-present], the effects team reportedly used AI to de-age a character quite effectively. This came on the heels of criticism over poor de-aging effects using more traditional techniques in The Mandalorian [source series of Boba Fett, Disney+, 2019-present].

Outside the entertainment industry, AI-generated audio can be used in digital assistant devices or to allow those who have lost their ability to speak to communicate in their own voice. It can even bring historic figures “back to life.”

In the last few years, the technology has advanced from requiring hours of processing time, hundreds of photographs or video samples, and a computer with reasonably advanced graphics capabilities, to something that can be created on your cell phone with a single selfie. Last August, reports surfaced about a tool that can even create deepfakes in real-time for streaming video. It’s likely that in the short time between the writing and publication of this article, additional technologies will be released.

Consent, consent, consent

Consent is particularly important in the context of digital humans, both in the original creation and any subsequent uses. The use cases described above require the consent of the person depicted, particularly in the case of professional performers, whose likenesses and voices are key to their livelihood. Civil Code Section 1708.86 provides a civil cause of action for non-consensual use of AI-generated sexual content. However, what about non-sexual content?

Last May, a Canadian voice actor sued ByteDance, Inc., the company behind TikTok, alleging that the company had used her voice in its text-to-speech tool. According to the complaint, Beverly Standing had performed voice work for a Scotland-based company “purportedly for Chinese translations,” and apparently without a written contract. The voice files were allegedly obtained by ByteDance and used for TikTok’s text-to-speech feature. Standing expressed concern that being branded as TikTok’s female voice, and being involuntarily associated with content she felt could reflect poorly on her image, would impact her ability to attract voiceover work, particularly for commercials, because she already would be associated with TikTok.

Standing sued on multiple theories relating to the unauthorized use of her voice, including violation of her right of publicity, false endorsement under the Lanham Act, and multiple other claims relating to unfair competition. There is precedent holding that the unauthorized use of an individual’s voice and/or digital likeness gives rise to at least a right of publicity claim, although the false endorsement and unfair competition claims might be more difficult in noncommercial uses.

Although this case ultimately settled, it illustrates some of the risks for both actors and content creators that are inherent in this space. More importantly, it illustrates the significance of clear contracts, particularly why it is important for creators to obtain consent and why performers need to carefully consider the scope of the consent.

Beware Granting Rights

An increasing number of reports and inquiries have been received by SAG-AFTRA from performers, agents, attorneys, and even other unions, regarding terms they are seeing in contracts for digital scanning and audio recording for AI-content. Even on traditional entertainment projects, performers are being asked to sign blanket releases granting rights to their voice, likeness, and performance well beyond the scope of what is necessary for the specific project.

Attorneys and others representing performers, models, or any other person for audio or performance capture should pay particularly close attention to these contract terms. This is as true if the contract is for traditional entertainment projects utilizing digital scanning techniques as it is for work specifically in the AI space.

Many of the companies currently working in the AI voice and video space are offering standard form contracts with very one-sided terms, often governed under the law of the foreign country in which the company is based. The contracts tend to have blanket grants of rights to use the performer or model’s likeness and voice irrevocably and in perpetuity. There is no approval right and no form of residuals or use-based payments. Some ask that the performer or model indemnify the company. Furthermore, some have morals clause with vague language.

Following are some questions to consider when reviewing clients’ contracts (both old and new) in light of these new innovations. Many of these questions seem innocuous or obvious, but they take on new meaning in this evolving space.

• What rights has the client granted or are being granted with regard to the use of the client’s voice or likeness? Does the grant of rights allow use beyond the current project? Do the intellectual property rights in the character allow use of voice or likeness in subsequent works?

It has always been important to have clear rights grants, including appropriate fences around use, in contracts that grant rights in a performer’s voice or likeness. The Standing case illustrates how much more important it is in the context of work using AI. It was largely the lack of a contract covering these terms that gave rise to multiple claims; an over-broad rights grant might have precluded them.

Actor contracts typically have a provision allowing the use of the performer’s likeness in character for purposes of merchandising. As expected, the producer retains all rights in the character. However, the increasing use of technologies that allow the creation of characters that are wholly digital, or even digitally enhanced, means attorneys need to pay closer attention to how likeness and voice rights grants are drafted, both in connection with the character and in merchandising.

The scope of any provision granting rights in a client’s voice or likeness, especially those relating to merchandising, is all the more important with the rapid growth of the metaverse and associated technologies. As the entertainment industry starts to experiment with nonfungible tokens—or, in common parlance, NFTs—attorneys should be sure their clients are protected from exploitation and have the right to control their likenesses in these new spaces.

These questions are of particular importance if the performer is being digitally scanned or providing likeness or voice for AI use, when digital assets are being created and will be owned and controlled by the producer. To the extent possible, counsel should seek to limit usage of the assets only to the current project, so they cannot simply be reused and repurposed without appropriate compensation.

Of note, if the work is done under a SAG-AFTRA agreement, SAG-AFTRA takes the position that any reuse of scanned content created for one project and used in a subsequent production falls within the reuse provision of the applicable agreement. SAG-AFTRA also aggressively objects to scanning contracts and other likeness and voice grants that extend beyond the individual project for which they are intended; with limited exceptions, the union’s agreements do not allow the producer to obtain reuse consent at the time the work is done and it must be separately bargained. These are important protections for performers, ensuring they are fairly compensated for their work and can control how their voice and likeness are used in the future.

• If a client is being scanned for motion or facial reference, such as for animation or video games, does it allow use beyond reference? For example, can the client’s likeness be used in the end product?

Actors, particularly those doing voiceover, are sometimes asked to do facial or other performance capture, ostensibly so that animators working on the project can capture their movements for character reference. Nevertheless, actors have reported that the use goes beyond simple reference to use of their likeness in connection with their characters. Performance capture technologies that are AI-driven can now do this by simply recording a client’s performance as it is delivered, without complicated rigging. Even if the employment contract is silent as to this point, counsel should look to ensure that a client is protected from this practice.

• Does the contract allow digital manipulation of a client, whether through AI or otherwise? Can the footage, itself, be manipulated even if the client cannot?

Digital manipulation goes beyond aging or de-aging a character, to things like depicting the person in a scene in which he or she did not perform, possibly even in a way the person would have objected to had he or she been present. California Civil Code Section 1708.86 provides a cause of action if that depiction involves nudity and sexual situations, and it would not apply to other manipulations. In the context of commercials and ads, in particular, if the producer has the right to alter a client or the footage, it opens up the possibility of creating completely new ads, potentially for a different product, without the client’s consent.

• Is there security around the audio and video recordings and the associated data? Do they have provisions in place to avoid unauthorized use or access? How is the content protected? What are the steps if there is a data breach?

Both the recordings and digital files of a client should be protected from unauthorized use and access, in the same manner as any personal information. Not only can an unauthorized use of a client’s voice and likeness harm the client’s brand or earning potential, in the hands of a malicious actor, the high-quality digital content can be used to create sexually explicit content, commit fraud, or spread disinformation. It is also worth determining if the content will contain any embedded technology that the producer can use to track content and if the producer will similarly assist the client in the event of unauthorized exploitation.

• If the project is an AI project, what control does a client have to approve or deny subsequent uses? Is the client comfortable with the fact that his or her voice could be used in an advertisement or other content for something the client might oppose?

These questions are particularly important for actors who work on commercials or who have other brand affiliations. Exclusivity is a critical component of many brand deals—an actor, spokesperson, influencer, or any other recognizable person cannot simultaneously be associated with, let alone be the face and/or voice of, two potentially conflicting brands. If a client lends his or her likeness or voice to a company that provides AI spokespeople or characters, without any approval rights or constraints on usage, the client risks foreclosing entire segments of future work.

Many of these platforms may engage talent to do this work early in their careers, often before they have engaged professional representation. They might be excited to work in this new space and not yet have an understanding about how this could impact their future job opportunities and earning capabilities, particularly if they are hoping to do commercials or have brand deals. The perpetual grant of rights, coupled with lack of control, is a significant risk for any performer but particularly for the performer just getting started.

• If the company does not allow project-by-project consent, does it at least have an ethics policy relating to how a client’s likeness or voice will be used? What rights does the client have if it changes?

Many of the companies developing these AI technologies come from the technology sector, rather than the entertainment sector, and they lack the types of protections or approval rights actors have come to expect. Many lack any approval rights at all. If the company lacks approval rights or will not negotiate them, it is important to determine if it at least has an ethics policy regarding how the content might be used and what rights a client might have to remove content from the company’s site should that policy change. While an ethics policy will not give a client control over his or her likeness or voice, it at least will give some reassurance as to how the client will be depicted.

• What indemnification has the client agreed to? What is the client receiving? Are there content carve-outs?

Unlike most acting work, a client likely will not have a copy of the script that will be used for a digital counterpart, and therefore will not be able to object to troubling, offensive, or even legally problematic content. One way to address this is to include carve-outs or consent requirements for certain types of content, such as profanity, sexual content, religious content, or the endorsement or advocacy for political positions or candidates. This can be coupled with indemnification for, at a minimum, content that is defamatory, casts the client in a false light, or is otherwise unlawful. At the same time, the indemnity the client is granting should be narrowly tailored and not extend to the content.

• Is the client a SAG-AFTRA member? If so, is the project signatory to a SAG-AFTRA agreement?

Voiceover and recorded performances in many of these new and evolving areas are within SAG-AFTRA’s jurisdiction, and SAG-AFTRA has been actively working to ensure its contracts keep up with technology. For the media professionals SAG-AFTRA represents, it is important to understand the implications of working on a non-union project in these areas. It not only risks running afoul of SAG-AFTRA’s Global Rule One but also means not having the union’s protections and support in the event of a dispute. SAG-AFTRA’s contracts in these areas recognize that these are evolving technologies and business models and have a degree of flexibility while still providing the types of minimum protections that have been discussed.

There is a rush to work in this exciting new space, but performers and models need to stop and think about the short- and long-term implications and risks that go with it. Is the upfront payment and excitement worth the potential long-term risk of overexposure or being associated with a product, company, or cause they do not support? Attorneys representing talent similarly need to be aware and keep apprised of the changing technology to properly advise their clients. As illustrated by Beverly Standing’s case, if a client does not maintain control over how his or her voice and likeness will be used, there is a risk it may be used in ways least expected.

[Danielle S. Van Lier is assistant general counsel for intellectual property and contracts at SAG-AFTRA in Los Angeles.]

Computers and Actors, Part 2

 

[This is the completion of “Computers and Actors,” a collection of articles about the increasing use of computer graphics and digital animation in film and even on the stage to enhance and even replace live actors.]

“ART MIMICS LIFE AND ANDY SERKIS”

[The article below was published in Screen Actor: Screen Actors Guild Magazine 45.3 (Summer 2004).  Screen Actor was a predecessor of SAG-AFTRA magazine, the current member publication of the union formed by the merger of the Screen Actors Guild (SAG) and the American Federation of Television and Radio Artists (AFTRA).  (Before the merger, the TV and radio union put out a magazine simply entitled AFTRA.)]

The story of Andy Serkis should leave every actor with two nuggets of inspiration: One, the notion that computer-generated  animation will soon eliminate the need for flesh-and-blood actors is nonsense. And two, the next time you’re offered a role that seems like it’ll limit your true talents, remember it could lead to artistic genius.

For Serkis, his journey to understanding these acting revelations started about four years ago when he answered a phone call from his agent. 

“Are you interested in doing three weeks of voice-over work for The Lord of the Rings in New Zealand?” the agent asked Serkis. “They want you to play a character named Gollum.”

The English-born actor hesitated.

“There must be 12 proper acting parts for that film,” Serkis said a bit sarcastically. “Can’t you get me something decent?”

It’s not like Serkis wasn’t intrigued or eager to work, but he was a veteran of stage and screen. He had a broad range of talents and skills to display. And despite reading The Hobbit by J.R.R. Tolkien, Serkis had never even cracked the Rings trilogy.

He voiced his concerns, but it was his wife who ultimately convinced him that inside that seemingly routine voice work was a rich, complicated character just waiting for life.

“So I met with Peter Jackson when he was casting in London,” Serkis said. “I was impressed right away.” Serkis admired the bearded director’s vision for the entire trilogy—and more specifically, Gollum. “He told me he wanted it to be the most actor-driven computer-generated character that’s ever been,” Serkis said. “He said we need to give him a heart and soul, so we’re going to use an actor.”

Jackson loved Serkis’ take on the tormented creature and offered him the job. But that still didn’t eliminate the actor’s apprehension. What exactly was he getting himself into?

“I was quite scared,” Serkis admits. “The notion of a computerized character did worry me.” Turns out, Serkis’ concerns were unwarranted. By now you know the rest of the tale. Three weeks would turn into four years as Serkis performed alongside Elijah Wood and Sean Astin—then found his likeness cropped right out of the film thanks to the magic of performance capture.

Although some may consider having your likeness erased from one of the most popular movie trilogies of all-time [sic] (except in [. . .] The Return of the King where he appears as Smeagol), Serkis says this job gave him a chance to perform some of the purest acting he’s ever done.

“It reminded me of the stage,” he says. “You still have to create the character. I still had to find that human emotion and that psychology.”

And what about the detractors who believe a character like Gollum may signal humans getting pushed out for pixels?

“Screen actors are manipulated all the time,” Serkis says. “The film is cut, it’s edited, music is added in behind.”

No matter what the medium, there must be a connection between actor and audience. And it’s no coincidence audiences made such a connection to Gollum. It’s because he is real after all. Thanks to Andy Serkis, who sums up performance capture technology like this: “I really think this is a fantastic new string in an actor’s bow.”

[Andy Serkis (b. 1964) played the performance-capture role of Gollum using motion capture acting, animation, and voice work for all three of the Lord of the Rings films (2001, 2002, and 2003) and in The Hobbit: An Unexpected Journey (2012).  He went on to appear as Kong in King Kong in 2005; Caesar in Rise of the Planet of the Apes (2011), Dawn of the Planet of the Apes (2014), and War for the Planet of the Apes (2017); Captain Haddock / Sir Francis Haddock in Steven Spielberg’s The Adventures of Tintin (2011); and Supreme Leader Snoke in Star Wars: Episode VII – The Force Awakens (2015).

[The movie website IMDb describes the technical work through which Serkis and the Lord of the Rings films’ creators crafted his character:

“Gollum” was the collaborative team’s effort around Serkis’s work in performance capture – an art form based on CGI-assisted acting. Serkis’s work was an interactive performance in a skin-tight CGI suit with markers allowing cameras to track and register 3D position for each marker. Serkis’ [sic] every nuance was picked up by several cameras positioned at precisely calculated angles to allow for the software to see enough information to process the image. The images of Serkis’ performances were translated into the digital format by animators at Weta Digital studio in New Zealand. There, his image was key-frame animated and then edited into the movie, Serkis did have one scene in “The Return of the King” showing how he originally had the ring, killing another hobbit to posses[s] it after they found it during a fishing trip. He drew from his three cats clearing fur balls out of their throats to develop the constricted voice he produced for “Gollum” and “Sméagol”, and it was also enhanced by sound editing in post-production.

Serkis spent almost two years in New Zealand and away from his family, and much of 2002 and 2003 in post-production studios for large periods of time, due to complexity of the creative process of bringing the character of “Gollum” to the screen. Serkis had to shoot two versions for every scene; one version was with him on camera, acting with (chiefly) Elijah Wood and Sean Astin, which served both to show Wood and Astin the moves so that they could precisely interact with the movements of “Gollum”, and to provide the CGI artists the subtleties of Gollum’s physical movements and facial expressions for their manual finishing of the animated images. In the other version, he’d [d]o the voice off-camera, as Wood and Astin repeated their movements as though “Gollum” were there with them; that take would be the basis for inserting the CGI Gollum used in the released movie. In post-production, Serkis was doing motion-capture wearing a skintight motion capture suit with CGI gear while acting as a virtual puppeteer redoing every single scene in the studio. Additional CGI rotomation [the technique in which animators trace live-action movement frame by frame] was done by animators using the human eye instead of the computer to capture the subtleties of Serkis’ performance. Serkis also used this art form in his performance as “Kong” in King Kong (2005), which won him a Toronto Film Critics Association Award (2005) for his unprecedented work helping to realize the main character in “King Kong”, and a Visual Effects Society Award (2006) for Outstanding Animated Character in a Live Action Motion Picture.

[Serkis has also been featured in his performance-capture personae in video games based on his films.  Before his success with performance-capture work, Serkis appeared in traditional acting roles on TV, in film, and on stage; since Lord of the Rings, he’s continued his career in those fields.  Serkis made his film directing début with 2018’s Mowgli: Legend of the Jungle, which included considerable performance-capture work and CGI.]

*  *  *  *
 “VIRTUAL SUPERSTARS?: THE PROMISE AND PERIL TO REAL ACTORS”
by Mark S. Lee
 

[Lee’s article about digitized actors replacing live ones was published in the same Summer 2004 issue of Screen Actor as the one above about Andy Serkis’s performance-capture work.]

In New Line’s Simone [see also “‘Movie Stars Fear Inroads By Upstart Digital Actors’” in “Computers and Actors,” Part 1, 4 October], a director used a digitally created “virtual” star to replace a “temperamental” actor. Marketed with the catchphrase “A Star Is . . . Created,” the film opened to less than stellar box office.

Although Simone was a fantasy, studios are working hard to make it real. New Line’s The Lord of the Rings: The Two Towers and The Return of the King prominently featured “Gollum” [see “‘Art Mimics Life and Andy Serkis,’” above], a digitally animated character created by special effects artists from the performance of a real actor. Universal’s Hulk [2003] starred a digitally created humanoid creature in the title role. DreamWorks recently entered into a “film sampling” agreement in which new story lines and actors would be inserted into existing motion pictures to create “entirely new entertainment.”

As these films demonstrate, digital technology creates both opportunities and dangers for real actors. An actor in theory could have her appearance digitally altered to make her appear 18 to perform in a high school comedy, or 25 to play a romantic lead, or 75 to play a character role, regardless of her real age. Height, weight, physique, all could be digitally manipulated. Actors might consider creating and retaining a library of their own digital images, poses, etc. for such purposes.

However, the same technology could make real actors obsolete. It could allow studios to create completely virtual “actors” such as Simone. Or, it could force actors to compete with themselves or deceased actors by allowing studios to superimpose images of famous living or deceased actors on an unknown who actually performs the role, or on photorealistic animated characters. New detective stories starring a virtual “Humphrey Bogart,” or new sex comedies starring a virtual “John F. Kennedy” and “Marilyn Monroe” could be in our future.

Surprisingly, it might be legal to make such films without the permission of the depicted actor or his or her estate. Unfair competition case law gives actors certain rights, but producers might avoid liability for such claims by simply providing a disclaimer. Defamation law would rarely apply to fictional works and could never protect the deceased. SAG’s Basic Agreement has a “reuse” provision that might protect actors when covered footage is used in the digitization process. SAG also has been heavily involved in strengthening state “right of publicity” laws, which prohibit unauthorized commercial exploitation of an individual’s name or likeness in certain circumstances. However, statutory or judicial exceptions to those laws involving fictional works would make recovery at best difficult, and at worst impossible.

A new federal law that protected actors’ ability to control how their image is used in digital “performances” is the best hope of protecting actors’ ability to pursue their livelihood. Absent such legislation, unauthorized uses of images in motion pictures, television programs, music videos, or Internet websites are likely to increase.

While much work remains to be done in this area, Screen Actors Guild has been successful covering certain “motion captured performances” through both direct bargaining with producers and through the arbitration process.

[Mark S. Lee is an intellectual property litigator in Los Angeles who has represented Tiger Woods; Elvis Presley Enterprises, Inc.; the Estate of Frank Sinatra; and others in unfair competition and right of publicity litigation.]

*  *  *  *
“DEAD STARS, ALIVE AGAIN; YES, MARILYN MAY FALL IN LOVE WITH VIGGO”
by Stuart Klawans
 

[Klawan’s “Dead Stars, Alive Again” was published in the “Week in Review” section (sec. 4) of the Sunday New York Times of 1 August 2004.]

THE visual arts, wrote the French critic André Bazin [1918-58] in 1945, have “a mummy complex” – they spring from a desire to preserve the human body from decay.

The power of film, Mr. Bazin argued, is that it embalms more completely than any previous medium, preserving not only anatomy but movements and sounds. We tend to think of film images as real, because the person who was preserved must have been present before the camera.

The funerary aspect of Bazin’s theory got a boost last week, but his case for realism took a dive, with reports on how computer technology has resurrected Sir Laurence Olivier, who died in 1989. A combination of manipulated archival film footage and fresh soundtrack dialogue will give Olivier a role – speaking lines he never spoke and making gestures he never made – in a new movie, “Sky Captain and the World of Tomorrow,” scheduled for release in September [14 September 2004 for the Hollywood premiere; 17 September for the rest of the U.S.].

“I haven’t seen that many movies with Olivier recently,” the movie’s producer, Jon Avnet, explained in an interview, “which is part of the reason why I thought this was pretty cool casting – that, and the fact that he’s a great actor.”

This is not the first time that a dead actor has been exhumed. For Carl Reiner’s 1982 comedy “Dead Men Don’t Wear Plaid,” special effects allowed Steve Martin to play scenes with Humphrey Bogart and James Cagney.

[A week after the publication of this article, the New York Times issued a correction of the statement above relative to the timing of the death of James Cagney: “While his image as a young man appeared in the 1982 comedy ‘Dead Men Don't Wear Plaid,’ he did not die until 1986.”  Bogart, however, had died in 1957.]

In 1991 Bogart and Cagney, and Louis Armstrong [1901-71] for good measure, appeared with the singer Elton John in a Coke commercial.

The dead stars were essentially cutouts, limited to a repertoire of existing filmed gestures. In the meantime, digital technology was evolving toward the creation of wholly digital actors, who could do anything they were told and would never demand a roomier trailer.

So far, digital characters have not fared well. The 2001 thriller “Final Fantasy: The Spirits Within,” with an all-digital cast, and Andrew Niccol’s satire “Simone” (2002), about the first virtual actor to become a celebrity, were box office disappointments [both films are discussed in Part 1 of this post].

But what if the technologies for reanimating dead stars and for creating digital actors could be merged? Researchers have been working toward that end since at least 1987, when Swiss scientists premiered “Rendezvous à Montréal,” a short film featuring digital versions of Bogart and Marilyn Monroe [1926-62].

The reanimation of Olivier in “Sky Captain” seems to be a further step in this direction.

When asked if he had freed his virtual Olivier from a set of prerecorded gestures, Kerry Conran, the writer-director of “Sky Captain,” said: “Our needs were very specific, by design. Within those limits, however, we did create a scene that Olivier never performed, so anything is possible.”

Mr. Avnet put it another way: “We created a different performance of sorts through motion capture, modeling and computer-generated imagery. Could you create a completely new performance with what we have today? It would be pretty demanding. But clearly, the future is coming fast.”

But the implications are anything but clear. “How do you protect imaginary humans? Can they have their own right of publicity?” asks Joseph J. Beard, a law professor at St. John’s University, an expert on copyright and trademark law, including the postmortem rights of digitized celebrities.

Mr. Beard notes that a patchwork of laws governs the use of dead actors’ images.

“California protects the images of celebrities for 70 years after death [see Astaire Celebrity Image Protection Act in “‘Movie Stars Fear Inroads By Upstart Digital Actors’” in Part 1], which is the same as under copyright law,” he said. “New York, on the other hand, does not have postmortem protection. So Bogart, who died as a resident of California, has state protection, and Cagney, who died as a domiciliary of New York, does not.”

The situation will only grow more complex, Mr. Beard said, if virtual actors gain artificial intelligence and become autonomous.

What if a virtual actor that looks and sounds like Olivier, and has a temperament to match, gets into a dispute with a rival Oliv[i]er simulation that is competing for the same roles? What if the interests of both these cyber-actors conflict with those of the rights holders to the original Olivier’s image?

The more pressing question for audiences may be the fate of the actor’s soul. We are accustomed to think of the greatness of a performer as an expression of individuality. But when Olivier is no longer captured on film but manufactured on the computer, perhaps in multiple versions, we lose the very thing that art was supposedly preserving: our point of contact with the irreplaceable, finite person.

For Mr. Avnet, that’s no problem. Noting that “Sky Captain” toys nostalgically with dated 1940’s Hollywood images of the future, he says: “What holds this film together is its celebration of movie history. To me, Olivier is perfect casting, because he embodies that history.”

[All these articles were published before the advent of deepfakes, the computer technology that allows someone in an existing image or video to be replaced with someone else’s likeness with sufficient exactitude to deceive viewers.  Wikipedia reports:

There has been speculation about deepfakes being used for creating digital actors for future films.  Digitally constructed/altered humans have already been used in films before, and deepfakes could contribute new developments in the near future.  Deepfake technology has already been used by fans to insert faces into existing films, such as the insertion of Harrison Ford’s young face onto Han Solo’s face in Solo: A Star Wars Story [2018], and techniques similar to those used by deepfakes were used for the acting of Princess Leia in Rogue One [2016].

[Combined with the techniques mentioned in the articles in “Computers and Acting,” deepfakery will only make it easier and more convincing to create virtual actors in the mages of deceased or aged celebrities, or simply to “cast” digitized—and compliant—avatars of living (and working) actors.  Disney, it seems, has already begun to work in this area—and I wouldn’t count on them to be reanimating Mickey and Donald!]