The Federal Communications Commission (FCC) has released a new notice of proposed rulemaking that would impose new requirements for autodialed marketing calls and texts that use artificial intelligence (AI) to generate content. This notice builds on the FCC’s January 2024 decision declaring that AI-generated content qualifies as an “artificial voice” under the Telephone Consumer Protection Act (TCPA), and therefore marketing calls and texts using AI-generated content require prior written consent.

The notice proposes to define an “AI-generated call” as “a call that uses any technology or tool to generate an artificial or prerecorded voice or a text using computational technology or other machine learning, including predictive algorithms, and large language models, to process natural language and produce voice or text content to communicate with a called party over an outbound telephone call.” This definition covers only the content of calls, so other uses of AI to generate calls – such as AI targeting of consumers to call – would not trigger the new rules. (AI-generated dialing could be treated as an autodialer under the TCPA in some cases, but is not part of this rulemaking.) The definition does, however, include the use of AI-generated content at any point in a call, such as in the ordering process. The notice does not address whether the use of AI to create a prerecorded message (as opposed to interactive content that is responsive to the consumer’s reactions) would fall within the definition.

If a marketer wanted to use AI-generated content in its robocalls or texts, the proposed rules would impose two new requirements. First, the marketer would need to obtain prior express written consent, not just to make robocalls and texts but also to use AI-generated content, through a “clear and conspicuous” disclosure. The notice asks whether that disclosure and consent should be incorporated into the basic consent to receive calls or should be separate.

Second, the notice proposes to require companies making AI-generated calls or texts to “clearly disclose” the use of AI at the beginning of each call or text. The notice asks whether callers should be required to use standard indicators as part of these disclosures, such as badges in texts or specific language in calls.

In addition to the proposed rules, the notice asks for comment on whether existing consents for robocalls, which do not include specific approval for AI-generated content, should be exempted and treated as covering AI content once the new rules are in effect. If the FCC does not allow existing consents, any company using AI-generated content will either have to make some calls without that content or obtain new consents from all the people it calls.

These rules would not apply to inbound calling, which is outside the scope of the TCPA. The FCC also asks for comment on potential exemptions from the AI-related requirements – including an exemption for the use of AI technologies by people with disabilities. Finally, the notice asks whether consumers should be permitted to opt out separately from AI-generated calls, while still permitting calls that do not use AI.

Comments will be due 30 days after the FCC publishes notice of the proposals in the Federal Register, and reply comments will be due 15 days after the comment deadline. It is likely that the new rules will not be adopted until 2025, and they may not go into effect for some period after they are adopted. The notice had bipartisan support, so it is likely that rules for AI-generated calls will be adopted regardless of the results of the presidential election.