Answer capsule
The CMO should approve an AI-avatar campaign only after deciding what audience claim the avatar conveys and whether any portrayed customer experience is real, supported, authorized, and governed through the full agency and platform chain.
What the source establishes
- The FTC's Consumer Reviews and Testimonials Rule took effect October 21, 2024, and the agency's Q&A says its staff guidance is not comprehensive and offers no safe harbor.
- The Q&A says the rule has no blanket prohibition on AI-generated stock avatars and that a stock avatar is not itself giving a consumer review.
- The FTC says an avatar could convey a testimonial, and the underlying testimonial would be prohibited by the rule if it were fake or false.
- The Q&A says a celebrity avatar used without permission can violate the rule when reasonable consumers would believe the celebrity gave the favorable testimonial.
Classify the message before approving the asset
The direct marketing answer is that an AI avatar is a production method, not a complete legal or audience classification. The accountable CMO should decide what a reasonable viewer is likely to understand: a fictional presenter, a dramatization, a customer, a paid endorser, a company representative, or a recognizable person. The same synthetic face can carry different implications depending on the script, first-person language, product demonstration, placement, caption, channel, and surrounding creative.
That decision needs the served version, not only a studio description. A label such as virtual spokesperson may help but cannot automatically reverse a script that claims personal use or results. Conversely, a clearly fictional character does not become a consumer merely because it speaks naturally. Marketing should preserve the creative, voice, identity source, script, claim, approval, disclosure, and intended audience so the classification rests on evidence rather than the novelty of the tool.
The accountable team should translate this point into a named workflow, affected population, source data, human owner, approval right, exception path, retained evidence, and review date. That translation is what separates an interesting AI development from a decision that can be governed and evaluated.
Trace any experience claim to a real source
The FTC distinguishes the avatar from the underlying testimonial. If the message represents that someone used a product, achieved a result, preferred a service, or holds an opinion based on experience, the CMO needs to know whose experience is being communicated and what supports it. Generating a believable presenter does not generate a customer history. A composite script, transformed quote, translated statement, or agency-written line can change the net impression even when it began with authentic material.
The approval record should separate identity authorization, experience evidence, claim substantiation, typicality, material connections, and disclosure. Those conclusions answer different questions. Permission to use a likeness does not establish the truth of a performance claim; a real customer's statement does not establish that the result is typical; and a disclosure does not cure a fabricated experience. Unsupported elements should remain unknown or be removed rather than blended into a general authenticity score.
The accountable team should translate this point into a named workflow, affected population, source data, human owner, approval right, exception path, retained evidence, and review date. That translation is what separates an interesting AI development from a decision that can be governed and evaluated.
Keep the agency and platform chain accountable
The FTC Q&A says advertising agencies, public-relations firms, review brokers, and reputation-management companies are not categorically immune from the rule. For a CMO, that means vendor creation does not move the campaign outside the brand's governance. The record should identify who supplied the script, avatar, voice, likeness permission, source testimonial, editing, translation, disclosure, placement, and final approval, together with the representations each party actually made.
Platform features can also change the audience experience. Cropping, autoplay, voice replacement, localization, caption placement, influencer reposting, or paid amplification may alter which disclosure appears and whether the content looks like a personal account. Marketing should review the material version delivered in each important channel and preserve evidence of how it appeared. A platform-generated AI marker can inform viewers without answering whether the underlying testimonial is truthful.
The accountable team should translate this point into a named workflow, affected population, source data, human owner, approval right, exception path, retained evidence, and review date. That translation is what separates an interesting AI development from a decision that can be governed and evaluated.
Separate the rule from broader deception analysis
FTC staff notes that some conduct involving actors or avatars may fall outside a specific rule provision and still be deceptive under the FTC Act. The CMO should not treat a narrow classification as a blanket clearance. The decision still needs the campaign's net impression, audience, claim support, identity rights, disclosures, and other applicable requirements. Qualified counsel should interpret the rule and current facts; the marketing owner remains responsible for the creative and distribution evidence.
The staff Q&A is useful because it rejects two shortcuts at once: AI avatars are not automatically prohibited, and they are not automatically harmless. The accountable decision is whether this exact message truthfully communicates who is speaking and what experience, if any, supports the claim. That conclusion can support approval, revision, a narrower channel, or rejection without turning the existence of synthetic production into either proof of deception or proof of compliance.
The accountable team should translate this point into a named workflow, affected population, source data, human owner, approval right, exception path, retained evidence, and review date. That translation is what separates an interesting AI development from a decision that can be governed and evaluated.
Decision test
Ask whether the source changes the decision itself, the evidence required, the implementation sequence, or only the language used to describe an existing capability. Record which claims are directly supported, which are provider statements, which require an independent test, and which remain unknown. A source-linked review should make uncertainty easier to see, not bury it inside a blended score.
Questions to take into review
- Can the asset's origin and edits be reconstructed?
- Which disclosures apply by market and context?
- What training, input, and output rights apply?
- Which review gates cover claims and brand expression?
- Which repository owns approved content?
- How are market and channel variations controlled?
- Which catalog and policy records ground answers?
- How are sponsored recommendations disclosed?
The publication supports research and executive decision preparation. It does not provide legal, financial, accounting, employment, clinical, cybersecurity, investment, procurement, or implementation advice.