The Consent Equation: When Synthetic Performances Force Hollywood to Define What a Performance Actually Owns

The conversation usually begins the same way. A director has a vision. A producer has a budget constraint. And somewhere in the middle of that negotiation, a VFX supervisor finds themselves being asked a question that has no clean answer: can we make them look like that actor?

In some cases, the actor in question is unavailable — committed to another project, unwilling to return for a reshoot, or no longer living. In others, the request is more speculative: a synthetic version of a performer crafted entirely from reference material, trained on decades of screen appearances, capable of delivering lines that were never spoken and inhabiting scenes that were never filmed. The technology to accomplish this exists. The consensus on whether it should be used does not.

This is the conversation Hollywood is actively avoiding, even as the decisions it refuses to make accumulate into precedent.

The Technical Threshold Has Already Been Crossed

For years, the ethical debate around synthetic celebrity performances existed at a comfortable remove from practical reality. The technology was impressive in demonstration but unconvincing in deployment — close enough to unsettle, not close enough to deceive. That distance has collapsed.

Contemporary deepfake and neural rendering pipelines, when applied by skilled VFX professionals with sufficient training data and production resources, are now capable of generating performances that pass casual scrutiny and, in controlled viewing conditions, challenge even careful examination. The uncanny valley that once provided a natural brake on synthetic performance deployment has been narrowed to the point where it no longer functions as a reliable deterrent.

VFX supervisors working at the intersection of this technology and live production describe a shift that has occurred not gradually but in discrete jumps. A technique that required months of work and produced marginal results three years ago now yields materially superior output in a fraction of the time. The acceleration has outpaced institutional response at every level — legal, contractual, regulatory, and ethical.

What Existing Contracts Actually Say

Entertainment attorneys who specialize in talent agreements are candid about the inadequacy of existing legal frameworks. Standard performer contracts, even those negotiated in recent years, were not drafted with the assumption that a studio might generate a wholly synthetic version of the signatory using machine learning models trained on their prior work.

Some agreements include likeness provisions that prohibit unauthorized digital reproduction. Others contain language broad enough to be interpreted as permitting it, depending on how terms like "likeness," "performance," and "derivative work" are defined — definitions that vary by jurisdiction, by contract vintage, and by the specific nature of the synthetic output in question.

The situation involving deceased performers is particularly fraught. Rights to a deceased performer's likeness typically vest with their estate, but the scope of those rights — and the extent to which they cover AI-generated synthetic performances — remains actively contested in American courts. Several high-profile cases are currently working through the system, and the outcomes will establish precedent that the industry will be living with for decades.

The Screen Actors Guild-American Federation of Television and Radio Artists secured provisions addressing AI and synthetic performance in its 2023 contract negotiations, but those provisions address a landscape that is already evolving beyond the scenarios they were designed to govern. The ink dries slowly in Hollywood; the technology does not.

The Professionals Caught in the Middle

For VFX artists, the ethical dimensions of synthetic performance work are not abstract. They are present in the daily execution of specific technical tasks — in the decision to push a digital likeness closer to photorealism, in the judgment calls made during facial reconstruction, in the awareness that the work being produced may be used in ways that the subject of that work never authorized.

Several VFX professionals, speaking with the understanding that their employers would prefer this conversation not occur publicly, describe a pattern of institutional pressure that discourages ethical inquiry. Questions about consent and authorization are redirected to legal departments. Legal departments provide assurances calibrated to minimize liability rather than to resolve genuine ambiguity. The work proceeds.

The discomfort is particularly acute in productions involving deceased performers. Reconstructing a living actor's performance involves a complex ethical calculus, but the subject of that reconstruction can, in principle, advocate for themselves. A performer who has died cannot object, cannot negotiate, cannot withdraw consent that was never explicitly given for applications that did not exist when they were alive.

Some artists have declined specific assignments on ethical grounds. That option, while available in principle, carries professional risk in an industry where reputation for reliability frequently outweighs reputation for principle.

The Conversations Studios Are Avoiding

The most significant ethical failures in this space are not occurring in moments of deliberate misconduct. They are occurring in the gaps between conversations that production entities are motivated not to initiate.

A studio that commissions a synthetic performance of a deceased performer without explicit estate authorization is making a legal and ethical bet — that the rights are ambiguous enough to defend, that the estate will not litigate aggressively, that the cultural response will be muted. Those bets have sometimes been correct. The fact that they have sometimes been correct has encouraged more bets.

The living performer scenario generates a different but equally troubling dynamic. When a studio creates a synthetic version of a performer — for marketing, for supplemental content, for scenes that the performer's schedule or compensation demands made impractical — it is making a claim about the separability of performance from performer. It is asserting that what makes an actor valuable can be extracted, replicated, and deployed independently of the actor's ongoing participation or consent.

That assertion, if it becomes normalized, has consequences that extend well beyond any individual production decision. It redefines what a performance is, what a performer owns, and what the industry's relationship to human creative labor actually means.

A Framework That Does Not Yet Exist

The professionals closest to this technology are not uniformly opposed to synthetic performance. Many recognize its legitimate applications — historical reconstructions, accessibility accommodations, creative extensions that performers would enthusiastically endorse. The objection is not to the capability. It is to the absence of a coherent framework for deploying it responsibly.

That framework will require contributions from legal scholars, labor representatives, ethicists, technologists, and the performers whose identities are at stake. It will require studios to accept constraints that reduce their operational flexibility. It will require the industry to have conversations it has been systematically deferring.

The technology is not waiting for that conversation to conclude. Neither, ultimately, is the audience.