How brands get consent for AI-generated people: the compliant workflow

By HyperKnown Team·Published September 1, 2026

Legal review is where AI ad campaigns stall. The creative is ready in days; the question nobody can answer is whose face that is and who agreed to what. This piece lays out the consent workflow that survives that review: what to collect, in what order, and which rules it satisfies. It is written for the marketer who has to get an AI-generated campaign past counsel, and for the counsel doing the asking.

What consent do brands need for AI-generated people?

The legal basis is the right of publicity: in most US states, commercial use of a person's identity requires their authorization, and unauthorized use creates liability that lands on whoever ran the ad. Federal law is moving the same way through the NO FAKES Act, which would add a national claim against unauthorized digital replicas. None of this restricts licensed use; it makes the license the thing that separates your campaign from an infringement claim.

Why can't you rely on a stock avatar library's terms?

Serious reviews stumble on the same three points. First, the consent chain is not yours: the library holds whatever rights it acquired from the original person, usually through a one-time buyout, and you are trusting that acquisition was clean without ever seeing it. Second, many avatar tools restrict paid advertising or specific claim categories in their license terms, which teams discover after the media plan is booked. Third, there is no per-campaign record: when a platform, regulator, or the person themselves asks who approved this specific ad, a subscription agreement with a software vendor is not an answer.

Licensing an identified real person inverts all three. The consent chain starts and ends with your campaign, the ad rights are stated in the license tier, and the record names the person, the brief, and the approval. The trade-offs between the two approaches are covered in real people vs synthetic avatars.

What does the compliant workflow look like?

1. Select verified talent. Work only with likenesses attached to an identity-verified person. Verification is what makes the later consent record mean something.
2. Write a scoped brief. State the campaign concept, the channels it will run on, the territory, the license duration, and any claims the presenter will make. Vague briefs produce vague consent.
3. Clear restrictions first. The brief should be screened against the person's recorded content restrictions before it reaches them. A brief that violates them should never be sent.
4. Get informed, specific approval. The person reviews your actual brief and approves or declines it. This approval, tied to the brief, is the consent that matters.
5. Issue the license and pay. The license document fixes the approved scope; payment confirms the exchange. Both should exist before production starts.
6. Produce within scope. Generate and edit only what the brief described. New concepts, new channels, or extended flight dates mean a new brief, not an assumption.
7. Label where disclosure applies. The EU AI Act requires realistic AI-generated depictions of people to be visibly disclosed, and major ad platforms have their own synthetic-media labeling rules. Build the label into the creative process rather than retrofitting it.
8. Retain the record. Keep the brief, the approval, the license, and the payment together, per campaign, for as long as the content can resurface.

What should the consent record contain?

Which rules are you satisfying?

• Right of publicity (state law). The license is your authorization; the record is your proof.
• Emerging digital-replica law. The NO FAKES Act and state statutes such as Tennessee's ELVIS Act penalize unauthorized replicas, which makes documented authorization the safe harbor in practice.
• EU AI Act disclosure. Article 50 of the AI Act requires deployers to visibly disclose AI media that could pass as authentic when distributed in the EU, on top of the machine-readable marking that providers embed.
• Platform synthetic-media policies. Major ad platforms require disclosure of AI-generated realistic people in various placements; a documented license also answers their misrepresentation reviews.
• Truth-in-advertising rules. Claim substantiation and endorsement rules apply to what the presenter says regardless of how the video was made. An AI presenter does not loosen them.
• Union agreements, where applicable. If the talent is a union performer, SAG-AFTRA's digital-replica provisions on consent and compensation apply on top of everything above.

How does this work on HyperKnown?

The workflow above is the product. Brands browse identity-verified creators, every one carrying a Blue Tick. Submitting a brief captures scope, and the brief is checked against the creator's recorded restrictions before delivery. The creator approves or declines; payment is collected only after approval, with 65% of the license fee allocated to the creator. The license tier defines the cleared channels, from social through digital ads to CTV and digital out-of-home, and every project leaves the consent record described here. Agencies with their own production pipeline can request license-only clearance and take the documented rights into their existing workflow.