The strikes of 2023 made AI a household labor issue, and the contracts that followed made one principle explicit: a performer’s face and voice are theirs, and using a digital replica requires consent and compensation. Three years on, the industry is discovering how hard that principle is to operationalize.

Where synthetic performance actually shows up

Forget wholesale AI movies — the real adoption is granular. Dialogue replacement without reshoots. De-aging and stunt-face compositing. Dubbing that matches an actor’s own voice across a dozen languages. Background performers generated instead of hired. Each use case sits at a different point on the consent-and-compensation spectrum, and each is being negotiated separately.

The estate economy

Deceased performers have become an asset class. Estates license likenesses for new work, and the legal patchwork governing post-mortem publicity rights — different in every jurisdiction — is straining under commercial pressure. Some performers now write digital-likeness terms into their wills the way they once handled royalties.

The disclosure question

Audiences say they want to know when performance is synthetic; studios worry labels will stigmatize routine post-production. Regulators in several markets are drafting disclosure requirements anyway. The likely settlement: credits that acknowledge digital replicas the way they acknowledge stunt doubles — normalized, standardized, and mostly unread.

  • Voice is the frontier: audio replicas are cheaper, better, and harder to detect than video.
  • Consent infrastructure — who tracks what a performer approved — is a growing business of its own.
  • Watch the first major lawsuit over training data containing performances; it will set the real precedent.

Every technology that threatened acting — sound, television, CGI — ended up expanding it. Whether AI follows that pattern depends on contracts being signed right now.