In February 2024 the US Federal Trade Commission issued a supplemental notice of proposed rulemaking on impersonation of government and businesses. The notice states that the proposed rule would cover AI-generated voice, images and other deepfakes, and links the rule to Section 5 of the FTC Act, which prohibits unfair or deceptive practices. For AI companies selling into the US or advertising there, the document puts brand protection, advertising compliance and platform governance on the same agenda.
What the rule targets
The proposal addresses two kinds of impersonation: of government agencies and of specific businesses. Deepfakes are singled out because voice cloning and image synthesis have made impersonation almost costless. A few seconds of synthetic audio can persuade a consumer that a company issued a notice or a promise. Section 5 supplies the legal footing: unfair or deceptive commercial conduct is already prohibited, and the rule would turn this category of impersonation from a case-by-case judgement into a clearer prohibition.
For a company, the exposure no longer comes only from its own advertising copy. It also comes from third parties synthesising its name, mark, voice and likeness.
Brand protection turns on evidence
When impersonating content appears, the first instinct is a platform complaint. Whether the complaint works depends on how complete the evidence is: page captures, publication time, account details, distribution path and the context in which consumers encountered it. Deepfakes are often deleted or altered quickly, so preservation has to happen within a short window.
Monitoring scope also needs adjusting. The traditional focus was e-commerce listings and ad creative; short-video, audio and livestream clip platforms now belong in the same net. The aim is not to remove every similar item but to identify the ones that make consumers mistake the source.
Advertising and the licensing chain
Marketing is the other high-risk area. AI-generated voices, virtual presenters or digital endorsers can cross both impersonation and false-advertising lines when consumers are led to believe that a real person or a real company is endorsing something. Where the licence chain is incomplete, risk arrives from both ends: the holder of the synthesised likeness and the company implied to have endorsed it.
Endorsement, voice and likeness agreements should settle three points: whether AI synthesis is permitted, the permitted uses and channels, and what happens to synthetic material after termination. Contracts with agencies, MCNs and asset suppliers should also carry prohibitions on deepfake use and an obligation to cooperate with takedowns. Such clauses tend to prove their worth only after a dispute.
Platform mechanics and enforcement
Platform handling of impersonation mostly depends on reports from rights holders or the impersonated entity. The completeness of the submission drives the speed of removal; screenshots alone usually mean several rounds of correspondence. Once the proposal is finalised, enforcement tools will be clearer, but the remedies available to companies remain platform complaints, trademark enforcement and contractual claims, all resting on the same factual record.
For Chinese companies expanding abroad, the higher the US exposure, the greater the chance of having their brand synthesised. Linking monitoring, evidence, complaints and contractual recovery into one internal process is more useful now than waiting for the final text.
See our IP protection service for brand and mark protection, AI infringement response for deepfake takedowns, and commercial contracts for licensing and indemnity drafting.
This article is based on the supplemental notice of proposed rulemaking issued by the US Federal Trade Commission in February 2024 and on public information. It is for reference only and does not constitute legal advice.