When an AI model is asked to generate a real person, claya decides whether it's allowed, before the video exists. Allowed requests carry a licence fee at the price the rights holder set. Everything else never renders.
A prompt references a registered likeness. The rail stops it at 5%, before compute. A licence is issued at the rights holder's price, settled, and only then does the render continue. One transaction record from request to payment.
Video models can now produce any real person, on demand, at near-zero cost. That makes compute cheap and human identity the scarce input. Today it is used without consent, without terms and without payment.
Takedowns run after the fact. Watermarks label a render that already exists. claya moves the check before compute is allocated: one mandate per person, queried by the model on every request, settled to the rights holder on every authorized use.
Not a studio · not a model · not a marketplaceForty-five seconds, one question: who gets paid when a model makes your face? Made in-house with registered claya talent and an original track. Sound on.
Rights are configured once, in business terms. claya compiles them into what each provider needs and resolves every request against the same record. Legal permission, technical decision and fee stay tied to one mandate version.
Talent and representative set the grant, exclusions, approvals and economics. Once, not per deal.
claya resolves each request against that record: clear it, block it, or route it for approval.
The connected provider applies identity controls before rendering. Refusals, not takedowns.
Each authorized use links back to its mandate for review, reporting and settlement.
The representative's view: who is available, who has demand, what is clearing, what was stopped, what earned. Only exceptions reach a person. Everything inside the mandate clears automatically.
The mandate is model-agnostic; providers are activated separately. One frontier provider already enforces registered identity protection. A second provider running the same mandate logic is the next proof.
Need: one mandate, one control layer, distribution across every model.
Need: cleared talent, one integration, one settlement rail.
Agencies, labels, estates and individual artists. One mandate, filed once, enforced at model level and settled per use.
One roster agreement instead of model-by-model contracting. Onboarding, approvals, reporting and settlement for every identity you represent, through one operating layer.