On August 7, 2024 the Federal Communications Commission (FCC) adopted a new Notice of Proposed Rule Making (NPRM) proposing regulations that prohibit the use of AI in automated dialing or artificial or pre-recorded voice calls absent the prior written consent of the call recipient, unless otherwise exempted by the FCC.  The action was taken under the authority of Section 227(b)(2)(B) of the Telephone Consumer Protection Act (TCPA) (47 U.S.C. § 227) which prohibits pre-recorded or artificial voice calls to any residential telephone line unless exempted by the FCC. The proposed rulemaking was prompted over continuing concern over robocalling and robotexting and the impacts that AI generated voice calling may have in fostering more fraud and abuse.  

The proposed rules are intended to work within the existing framework of the “Telemarketing, Telephone Solicitation, and Facsimile Advertising” rules that prescribe limitations and various requirements upon telephone solicitations. Under the current regime, outbound telemarketing calling to telephone numbers listed in the National Do Not Call Registry is prohibited in the absence of the prior written consent of the consumer.[1] As proposed, the new FCC AI rules would operate within that framework by imposing clear disclosure and transparency requirements when obtaining consent, and attempts to address the issue of AI impersonation and false identity.[2]

As a first matter, the FCC proposes to adopt a definition of an “AI-generated call.”  The proposed definition covers “any technology or tool” that generates an artificial voice, prerecorded voice or text using “computational technology or machine learning.” The FCC further elucidates the definition by including “predictive algorithms and LLMs…to process natural language and communicate” with a called party.  Notably, within the definition calls are limited to “outbound calls,” thus interactive AI customer service, information retrieval, transaction processing, and other in-bound call-initiated functions fall outside the reach of the new rules.     

Addressing the substantive aspects of the proposed rules, the FCC seeks to impose clear disclosure obligations upon outbound telemarketing activities. Under the new rules, outbound marketers obtaining written consent to telemarket to a consumer must clearly and conspicuously state whether the permission includes permission to use AI-generated communications. This requirement would extend to text-based marketing solicitations as well. Finally, addressing antifraud, the FCC rules would require that any outbound call employing an AI-generated voice must disclose it is an artificial voice at the beginning of a call. As proposed, the AI voice disclosure would be included as part of the existing requirement that a caller’s identity be disclosed. Whether these new rules are sufficiently robust to meaningfully prevent AI generated robocalling and deceptive impersonation abuses is unclear. Though past history indicates it is unlikely absent advances in antifraud technologies. To this point, the NPRM includes a notice of inquiry (NOI) seeking comment about real-time call detection, call alerting, and call blocking technologies, including AI technologies capable of real time detection of objectionable or fraudulent content and patterns.  In its discussion, the FCC takes notice of Google’s call monitoring technology utilizing Gemini-Nano, which is a compact LLM designed for edge devices that could be deployed theoretically onto or coupled to smart phones devices. It also takes note of OriginStory, a technology that seeks to authenticate voices and establish a trust chain, and Microsoft’s Azure Operator Call Protection, which seeks to perform AI-assisted real time fraud detection at a carrier network level. Despite their promise, the FCC also recognized that these potential antifraud technologies raised concerns over user privacy, especially where content inspection is required.  

Interested parties may submit comments to the FCC within 30 days of the publication of the NPRM in the Federal Register. If you have questions about the NPRM or what it may mean for your operations, please contact the author.    

[1] Numbers that are not listed in the National Do Not Call Registry are not subject to federal calling restrictions.  However, states such as Connecticut have implemented their own mini-TCPA consumer protection laws prohibiting unsolicited telemarketing calls without prior written consent whether or not listed in the national registry upon a consumer’s voice directive to a marketer   See, An Act Concerning Charitable Organizations,

Telecommunications and The Attorney General’s Recommendations Regarding Consumer Protection, Conn. P.A. No. 23-98, (Eff. Oct. 1, 2023)

[2] The FCC previously determined that artificial intelligence voice calls fell within the scope of the phrase “artificial or prerecorded voice” under 47 U.S.C. § 227(b)(2)(B) in a declaratory ruling.  See, Implications of Artificial Intelligence Technologies on Protecting Consumers from Unwanted Robocalls and Robotexts, CG Docket No. 23-362, Declaratory Ruling, FCC 24-17 (rel. Feb. 8, 2024).

[View source.]