Tennessee Transparency for Deepfakes in Political Advertising Act (Tenn. Code Ann. § 2-19-120(a)(6)): AI Compliance Requirements

For any political communication in AUDIO form that is a deepfake impersonating a candidate for elected office in speech the candidate did not in fact engage in, state clearly at the beginning AND at the end of the communication: "This communication contains fake content impersonating speech that did not occur." The wording is prescribed by statute — this is not a general "AI-generated" labelling duty that you may paraphrase, and a single disclaimer at one end of the spot does not satisfy it. Build the double placement into radio, podcast, streaming-audio and robocall production templates, and treat synthetic-voice impersonation of a candidate as the trigger regardless of how brief the clip is.

For any political communication in VIDEO form that is a deepfake depicting a candidate for elected office in action or speech the candidate did not in fact engage in, display the prescribed statement in large letters FOR THE DURATION of the video — not as an opening or closing card. The required wording depends on what the communication contains: where it includes only video media, "This communication contains fake content depicting actions that did not occur"; where it includes both audio and video media of a candidate engaging in action or speech the candidate did not engage in, "This communication contains fake content depicting actions and speech that did not occur." Because the disclaimer must persist and be in large letters, it is a layout constraint on the creative rather than a post-production addendum: reserve the space in the template and verify the burn-in survives every aspect-ratio and platform re-encode.

Platforms and distributors are shielded only where they did not create the deepfake, and the shield is drawn in two distinct pieces — check which one you are relying on. Under § 2-19-120(a)(6)(D)(1) a host, third-party producer, or internet service provider including affiliates and subsidiaries, or an operator of a search engine or cloud service, is not liable solely for providing access or connection to a website or other information or content on the internet, nor is a facility, system or network not under the ISP's control, to the extent the provider is not responsible for creating the deceptive and fraudulent deepfake. Separately, § 2-19-120(a)(6)(D)(2) disapplies the subdivision entirely to a radio or television broadcasting station, cable or satellite television operator, or streaming or other programming service when it is PAID to broadcast material containing a deepfake, again to the extent it is not responsible for creation. Advertising agencies and production houses that generate or commission the synthetic content sit outside both shields and should not assume platform-style protection.