A regulator ruled that an AI voice on a robocall is still a voice

Start here

If a robocall uses a synthetic voice instead of a recording of a person actually speaking, is it still covered by the law limiting unwanted automated calls. On 8 February 2024 the Federal Communications Commission answered that directly. A Declaratory Ruling confirmed that calls made with AI-generated voices count as using an 'artificial or prerecorded voice' under the Telephone Consumer Protection Act, the 1991 law the FCC already used to limit junk calls. For a person receiving a call, the practical result is that the caller needed the same prior consent already required for any prerecorded robocall; the technology used to generate the voice creates no loophole.

What the documents say

The ruling itself, filed under CG Docket No. 23-362, states that 'the TCPA's restrictions on the use of artificial or prerecorded voice encompass current AI technologies that generate human voices,' citing a 2023 appeals decision, Trim v. Reward Zone USA, which described an artificial voice as 'a sound resembling a human voice that is originated by artificial intelligence.' The ruling followed a November 2023 Notice of Inquiry and drew support from 26 state attorneys general. The Commission's news release quotes then-Chairwoman Jessica Rosenworcel stating that bad actors used AI voices to 'extort vulnerable family members, imitate celebrities, and misinform voters,' and frames the ruling as giving state attorneys general new tools against the callers themselves, not just the resulting fraud. Neither document claims the ruling stops such calls outright; it changes what is enforceable and who can act.

Check this

The mechanism to notice is consent, not detection. The rule does not ask a listener to identify whether a voice was AI-generated; it requires the caller to have obtained prior express consent before using such a voice, under the same exceptions and written-consent rules that already applied to prerecorded messages. A reader can check whether a call was solicited: did you ever agree to receive automated calls from that number. If not, the call was likely already outside what the TCPA permits, and can be reported to the FCC's complaint center.

What holds and what fails

The ruling holds as a matter of legal classification: it closes the argument that an AI-generated voice sits outside 'artificial or prerecorded voice' language written in 1991. It does not hold as a technical stopper, since enforcement still depends on someone identifying and pursuing the caller, who may operate from outside reachable jurisdiction. Nor does it tell a listener how to detect an AI voice in real time; that remains a separate, unresolved problem.

• Do not assume a call is legitimate because the voice sounds human and specific.
• Check whether you gave prior consent to receive automated calls from that number.
• File a complaint with the FCC if you believe a call used a synthetic voice without consent.

The ruling's real work is legal plumbing: it keeps a decades-old consumer protection connected to a new way of generating the voice on the other end of the line, rather than creating a new rule from scratch.