Hollywood has spent years asking what artificial intelligence will do to entertainment, but the question is becoming much more immediate: what happens when the computer does not simply help make the movie, but performs in it? That distinction is at the center of one of the biggest conversations shaping the entertainment industry in 2026. Artificial intelligence can now generate convincing faces, voices, movements and performances, making it increasingly possible to create a performer who never existed, digitally reproduce a performer who does, or alter an existing actor’s performance without audiences immediately realizing what they are seeing. The technology is advancing faster than the entertainment industry’s rules can comfortably keep up, forcing Hollywood to reconsider what it means to own a performance, give consent to a performance and even be considered a performer in the first place. At the same time, governments are beginning to step in. New York has established disclosure requirements for advertisements that use synthetic performers, while SAG-AFTRA’s newly ratified 2026 television and theatrical agreement gives performers stronger protections surrounding digital replicas and wholly synthetic performers.

The idea of a “digital double” is not entirely new. Hollywood has been digitally altering actors for decades. Visual-effects teams have made performers look younger, older, thinner, larger, injured or physically transformed. Digital doubles have also been used for dangerous stunts, crowd scenes and sequences where it would be impossible or impractical to place an actor physically in every frame. What is changing now is the sophistication of the technology and the degree of independence it can provide. Instead of simply modifying an actor’s recorded performance, AI systems can generate entirely new material. A performer can potentially say words they never recorded, appear in a scene they never filmed or perform an action they never physically performed. The technology is moving the industry from digital assistance toward digital authorship.

That creates a fundamental problem: who actually owns the performance?

If an actor records a scene and a studio uses artificial intelligence to generate additional dialogue in that actor’s voice, the situation is relatively straightforward compared with a completely synthetic character. The actor exists, the original performance exists and the digital replica is derived from that person’s identifiable characteristics. But what happens when an AI-generated character looks like a human being but does not correspond to any specific actor? SAG-AFTRA makes an important distinction between these two categories. A “digital replica” is a digital asset that resembles a specific person, while a “synthetic” is an artificial digital asset that does not resemble a recognizable individual. Under the union’s 2026 agreement, both categories now receive significant protections, although the rules governing them are different.

That distinction may sound technical, but it could shape the future of casting.

Imagine a studio needs a background character for a television series. Instead of hiring an actor, it generates a synthetic person using AI. The character does not look like any particular real-world performer, so there is no obvious individual whose likeness has been copied. From the studio’s perspective, this could appear efficient. There is no scheduling conflict, no travel expense and potentially no need for repeated filming. The character can theoretically be modified whenever the production requires it.

But from the perspective of working performers, the problem is obvious.

If studios can replace enough small human roles with synthetic characters, the number of opportunities available to actors could gradually shrink. A single synthetic performer might not appear threatening. Thousands of synthetic performers could fundamentally change the economics of the industry. This is why SAG-AFTRA’s 2026 agreement places a strong emphasis on preserving human performance. Under the new agreement, producers must bargain with the union before using synthetics, and they have committed to a principle favoring human performances. They also agreed that they do not intend to use a synthetic in a role that would otherwise be performed by a human unless the synthetic provides “significant additional value” to the production.

That is a significant development.

The entertainment industry is not simply trying to ban AI. It is trying to establish boundaries around how AI can be used.

That distinction is important because artificial intelligence is already embedded in many parts of filmmaking. It can assist with visual effects, previsualization, editing, localization, dubbing, concept development, production planning and other technical processes. The issue becomes much more complicated when AI begins to replace the person whose performance is supposed to be at the center of the work.

Actors are therefore asking a very simple question: if my face, voice or performance can be turned into a digital asset, what rights do I have over that asset?

The answer has historically been fragmented.

Different states have different laws. Different contracts provide different protections. A person’s image, voice, copyright interests and publicity rights may be governed by different legal frameworks. This patchwork becomes increasingly difficult to navigate as technology crosses state and national borders. A digital replica can be created in one location, used by a production company in another and distributed globally within seconds.

That is why New York’s approach is particularly interesting.

New York has enacted a law requiring certain advertisements using synthetic performers to disclose that artificial intelligence was used to create them. The statute defines a synthetic performer as a digitally created asset generated or modified using generative AI or software that is intended to create the impression of a human audiovisual or visual performance without being recognizable as an identifiable natural performer. The law requires disclosure in covered advertisements, creating a clear principle: consumers should not be misled about whether the performer they are seeing is real.

The significance of that rule goes beyond advertising.

It establishes an increasingly important cultural expectation: transparency.

Audiences may be willing to watch AI-generated performances, but they should know what they are watching.

That principle could become one of the defining rules of the AI entertainment era.

For decades, audiences have accepted special effects because they understand that movies contain illusions. A dinosaur does not need to be real for Jurassic Park to work. A spaceship does not need to exist for a science-fiction film to feel convincing. But there is a difference between creating a fictional creature and creating a fictional human being who appears to be a real performer. Human faces and voices carry identity. They represent actual people, experiences and careers. The more convincing AI becomes, the more important that distinction becomes.

The controversy surrounding AI-generated performers has already demonstrated how sensitive this issue can be.

In 2025, SAG-AFTRA publicly criticized the creation of “Tilly Norwood,” an AI-generated actor-like character promoted as a potential new performer. The union argued that a computer-generated character should not be treated as an actor and expressed concern about AI systems being trained on performers’ work without permission or compensation. The debate surrounding Tilly Norwood became an early warning sign of how quickly synthetic performers could move from an experimental concept into a serious labor issue.

The question was not simply whether Tilly looked convincing.

It was whether anyone should be able to create a performer from technology trained on the work of real performers and then compete against those same people for jobs.

That question remains unresolved at the broader industry level.

Hollywood’s writers, actors and directors have all been forced to consider different versions of the AI problem. Actors are concerned about digital replicas and synthetic performances. Writers are concerned about AI-generated scripts and training data. Directors face questions about authorship and creative control. Behind all of those debates is one larger issue: if artificial intelligence can reproduce elements of creative work, how much of the human contribution remains protected?

The 2026 SAG-AFTRA agreement represents an attempt to answer at least part of that question for performers.

The agreement includes protections covering digital replicas, synthetics, biometric data, digital alterations and other forms of AI use. It also requires an articulable business reason for scanning performers, strengthening the principle that a performer should not simply be turned into a permanent digital asset because a production company wants the option.

That last point could prove especially important.

Scanning a performer is not the same as taking a photograph.

A scan can potentially capture detailed information about a person’s face and body that can later be manipulated to create performances the actor never actually gave. Once that data exists, questions about storage, security, ownership and future use become critical. The 2026 agreement specifically addresses the security of digital replicas and states that those protections should survive changes in ownership.

In other words, the performer is not simply negotiating for today’s movie.

They are negotiating for the future life of their digital identity.

That may become one of the most valuable assets an actor owns.

Consider a major star whose face and voice have become recognizable worldwide. Traditionally, that actor’s earning power depends on their ability to appear in films, television shows, advertisements and other projects. In an AI-driven entertainment economy, their digital identity could potentially become a separate commercial asset. A studio could theoretically want to use the actor’s digital replica after the actor has finished filming. A video game company might want the actor’s voice. An advertising agency might want a digital version of the actor to appear in a campaign.

The actor therefore needs more than a paycheck.

They need control.

Consent becomes the foundation of that control.

SAG-AFTRA has repeatedly emphasized consent and bargaining when it comes to digital replicas. Its current AI resources state that the 2026 agreement expands protections against unauthorized replication and gives performers additional safeguards against AI replacing their work.

The concept of consent becomes even more complicated when the performer is deceased.

Hollywood has already begun exploring digital recreations of deceased celebrities. One recent example involves the digital recreation of Gene Wilder as Willy Wonka for Netflix’s upcoming project “Wonka’s The Golden Ticket,” using AI-generated voice technology with the cooperation of Wilder’s widow. The development has generated debate about whether digitally reviving deceased stars is a tribute, a business opportunity or both.

SAG-AFTRA addressed another such case in March 2026 when it discussed the digital replication of Val Kilmer. The union stated that the applicable agreement and state law require consent from the estate for the use of a deceased performer’s digital replica in a motion picture, and said reporting indicated that Kilmer’s family had provided that consent. The union emphasized that digital-replica use must be transparent, authorized and consistent with contractual and legal requirements.

This introduces an entirely new question about celebrity legacy.

Should a person be able to decide what happens to their digital identity after death?

Should their family control it?

Should their estate be allowed to license it indefinitely?

What if the person never wanted to be digitally recreated?

What if the estate wants the money but the public sees the recreation as inappropriate?

And what happens when the digital version becomes more commercially valuable than the original recordings?

These are no longer science-fiction questions.

They are becoming business questions.

The economic potential is enormous. A digital performer does not get tired. A synthetic character does not age. A digital replica can potentially appear in multiple places. A voice can be localized into different languages. A face can be adapted to different scenes. A character can theoretically be available whenever a production needs it.

From a studio perspective, those possibilities are difficult to ignore.

From a performer perspective, they can feel like a threat.

That tension explains why Hollywood’s AI debate is not simply about whether technology is good or bad.

It is about who controls it.

A camera can capture an actor.

AI can potentially reproduce the actor.

The question is who gets to decide what happens next.

New York’s advertising disclosure requirement approaches the issue from the audience’s perspective. SAG-AFTRA’s contract approaches it from the worker’s perspective. Together, they point toward a broader framework that could eventually become standard across entertainment: disclosure, consent and compensation.

Audiences may eventually become accustomed to labels indicating whether a performance is human, digitally altered, digitally replicated or entirely synthetic.

That could become as normal as content warnings, ratings and credits.

Imagine opening a film and seeing a clear disclosure that a supporting character is entirely synthetic. Imagine an advertisement stating that the spokesperson’s appearance and voice were generated digitally. Imagine a streaming series explaining that an actor’s younger appearance was created using a licensed digital replica.

At first, those disclosures might feel strange.

Eventually, they may become routine.

And routine transparency could actually help the technology.

If audiences know what they are seeing, they can judge the creative work on its own terms.

The biggest danger for Hollywood may not be AI itself.

It may be deception.

If audiences believe they are watching a real actor when they are actually watching a synthetic performance, trust begins to disappear. The same is true for advertising. If consumers believe a celebrity personally endorsed something when a computer created the celebrity’s likeness, the difference is not merely technical. It changes the meaning of the endorsement.

This is why the New York law is so significant. It treats synthetic performers not simply as a technological novelty but as something that can affect how consumers interpret commercial messages.

The entertainment industry is therefore entering an era where the phrase “starring” may need a new definition.

Does a movie star have to be human?

Does a performance have to be physically performed?

Can an actor license their younger self?

Can an estate license a dead celebrity?

Can a studio create a fictional performer from scratch?

Can audiences develop emotional attachments to characters that were never performed by a person?

There may not be one answer to all of those questions.

And perhaps there should not be.

AI could genuinely expand creative possibilities. It could allow filmmakers to create characters that would be impossible to cast traditionally. It could help independent productions work with smaller budgets. It could make visual effects more accessible. It could assist performers rather than replace them. It could allow actors to perform across languages without recording every line again. It could help preserve elements of performances for historical or artistic purposes.

The technology itself is not inherently the villain.

The rules surrounding it are what will determine whether it becomes a creative tool or a mechanism for exploitation.

That is why the new SAG-AFTRA agreement is so important. Rather than attempting to pretend that synthetic technology does not exist, the agreement recognizes that it is becoming part of the production landscape and establishes conditions around its use. The union’s position strongly favors human performance while allowing room for synthetic technology when it offers significant additional value.

That approach may eventually become the industry’s middle ground.

Human first.

Technology second.

Consent always.

The next major challenge will be enforcement.

Technology moves globally, while contracts and laws are often geographically limited. A performer may have protections under a union agreement in one production but encounter a completely different legal environment in another country. A digital replica can cross borders instantly. A voice model can be hosted on a server halfway around the world. A synthetic advertisement can be distributed globally within minutes.

That means Hollywood’s AI rules will likely continue evolving.

New York’s disclosure requirement may inspire similar legislation elsewhere. California has already been strengthening protections around digital replicas, while federal proposals such as the NO FAKES Act continue to attract attention. SAG-AFTRA’s AI resources note that the NO FAKES Act was reintroduced in Congress in May 2026 with support from labor organizations and major technology and entertainment stakeholders.

The legal landscape is therefore moving, but it is not settled.

And that uncertainty may be the defining feature of the next few years.

Hollywood is accustomed to technological disruption. Sound changed movies. Color changed movies. Television changed movies. Digital cameras changed production. Streaming changed distribution. Visual effects transformed what filmmakers could put on screen.

AI may be different because it does not merely change how entertainment is produced.

It can potentially change who produces it.

That is why the phrase “digital double” feels increasingly inadequate.

A digital double once meant a tool used to extend a human performance.

Now, the digital performer can potentially become the performance.

That is the line Hollywood is trying to draw.

The future may not be a world where human actors disappear and computers take every role. It may instead become a hybrid industry where human performers and synthetic characters coexist. Some roles may remain completely human. Others may involve licensed digital replicas. Some fictional characters may be entirely synthetic. Visual effects may continue to blend human and computer-generated performances so seamlessly that audiences rarely think about where one ends and the other begins.

The important thing is that viewers should know.

Performers should consent.

Creators should be compensated.

And studios should have clear rules.

Because the most valuable thing in entertainment has never really been the technology.

It has been the human connection.

A great performance works because an audience believes that someone is feeling something. A voice makes a character memorable because it carries personality. A face matters because viewers associate it with a person, a history and a body of work.

AI can imitate those things.

It may eventually become exceptionally good at imitating them.

But imitation and experience are not necessarily the same thing.

That is the debate Hollywood is now entering.

The industry does not have to reject artificial intelligence to protect human creativity. It has to decide where technology serves the performer and where it begins replacing the performer. It has to determine when a digital replica is an extension of someone’s work and when it becomes a separate commercial performance. It has to establish what happens to someone’s digital identity after death. And it has to make sure audiences are not unknowingly watching a computer-generated person while believing they are watching a human being.

The answers will not come from Hollywood alone.

They will come from unions, lawmakers, studios, technology companies, performers and audiences.

And audiences may ultimately have the strongest influence of all.

If people demand transparency, studios will have an incentive to provide it.

If viewers prefer human performances, that preference will matter.

If audiences embrace synthetic characters, the industry will expand them.

The future of the digital double will therefore be shaped not only in boardrooms and legislative chambers, but also by what people choose to watch.

Hollywood is standing at a strange new threshold.

The computer can already create the face.

It can generate the voice.

It can reproduce the movement.

It can construct the scene.

Now the industry is deciding whether it should be allowed to perform the part.

The answer will determine much more than the future of acting.

It could determine what audiences mean when they say they watched a movie, what performers mean when they say they own their work, and what Hollywood means when it says someone is starring in a film.

The digital double has arrived.

Now Hollywood has to decide who gets to control it.