Hollywood briefing / AI likeness rights

AI Likeness Rights and the New Hollywood Contract Language

AI likeness rights are becoming less abstract because the contract now has to describe the face, the voice, the file, and the future use.

AI likeness rightsAI voice rightsAI actor contractdigital likeness consentsynthetic performer rights

AI likeness rights describe a person's control over AI-generated uses of their face, voice, name, gestures, or recognizable performance identity. In Hollywood, that means the contract has to catch up with the machine. A clause that once covered a publicity still may not be enough for a model that can keep speaking.

The new contract language is not glamorous. It is duration, territory, purpose, approval, compensation, revocation, estate control, and whether the file can be reused outside the original work. But that language is where the creative future gets decided.

A likeness is no longer only an image

For most of screen history, likeness meant photographs, posters, interviews, clips, merchandising, and publicity. AI changes the object. A likeness can now behave. It can read new lines, age down, sing in another language, perform inside a game, appear in an ad, or sit inside a database waiting for a future producer to discover it.

That shift makes old permissions feel thin. If a performer agreed to be photographed, did they agree to be simulated? If a singer licensed a recording, did they license a voice model? If an actor approved a scan for one film, did they approve the same body for a sequel, a trailer, or a brand campaign?

The words producers should expect to see

The language will keep narrowing. Consent must say what is being created. Compensation must say when money is owed. Approval must say whether the performer can refuse specific uses. Duration must say when the license ends. Scope must say whether the replica can appear in film, streaming, games, advertising, localization, archival restoration, or promotional material.

Estate control will become its own negotiation. The likeness of a dead performer can carry affection, grief, and commercial power. It can also carry risk if heirs, audiences, or guilds believe the use turns tribute into extraction.

The creative risk is moral confusion

A production can clear a use and still make the audience uncomfortable. That is the harder problem. Viewers may accept de-aging, dubbing, or stunt support when the performer is visibly protected. They may reject a synthetic performance if it looks like the person has been made to work after losing the ability to say no.

That is why the contract and the story cannot be separated. The rights language answers whether the use is allowed. The story answers whether the use feels earned.

Why this belongs beside book-to-screen IP

Book-to-screen properties give producers something AI cannot manufacture on demand: a tested emotional architecture. A novel can carry motive, atmosphere, memory, character pressure, and the moral terms of the world before a face is attached. In the AI likeness era, that matters. The story can tell the audience why the image exists.

Cassie Hour makes that tension visible. Its central object is a song, a voice, and a remembered woman whose image has been handled by other people. The novel is not a legal explainer. It is the human pressure under the contract language.

The signal to watch

Watch for deal terms around approval rights, synthetic dialogue, posthumous uses, localization, and promotional replicas. Watch for talent agencies building protected asset systems. Watch for unions pressing for clearer language. Then watch which producers understand that a clean likeness file still needs a story worthy of the face.

Sources and signals: SAG-AFTRA AI protections overview, Congress.gov NO FAKES Act listing, U.S. Copyright Office digital replicas report. This briefing is editorial analysis, not legal advice.