AI Voices and the TCPA: What US Law Actually Requires for AI Phone Calls
In January 2024, an AI-cloned voice of President Biden told New Hampshire voters not to vote. Weeks later, the FCC settled a question that had been ambiguous for years: does a decades-old anti-robocall law cover AI-generated voices? A sourced walkthrough of the ruling, the penalties that followed, and what it actually means for any business using AI to make outbound calls.
What the FCC actually did, and when
Feb 8, 2024
FCC Declaratory Ruling 24-17 confirmed the TCPA's "artificial or prerecorded voice" ban already covers AI-generated and AI-cloned voices.
NH primary trigger
An AI-cloned Biden robocall told voters not to vote, two days before the Jan 23, 2024 New Hampshire Democratic primary.
$6M + $1M
Proposed FCC fine against consultant Steve Kramer, plus a separate $1M settlement with carrier Lingo Telecom.
Disclosure rule: pending
A 2024 FCC proposal to require AI-voice disclosure at the start of every call had not been finalized as of the most recent tracking available.
The ruling: AI voices are "artificial" voices under existing law
The Telephone Consumer Protection Act has banned calls using an "artificial or prerecorded voice" without consent since 1991 -- decades before generative AI made voice cloning cheap and accessible. On February 8, 2024, the FCC adopted Declaratory Ruling FCC 24-17 (Docket CG 23-362), confirming that this existing statutory language already covers AI-generated and AI-cloned voices. Legally, this was an interpretation of a law already on the books, not a new rule requiring its own rulemaking process -- which is part of why the FCC could act on it quickly.
The practical effect: any outbound call using an AI-generated voice is now treated exactly like a traditional robocall under the TCPA. That means it requires the called party's prior express consent, is subject to existing disclosure and opt-out requirements, and exposes callers to TCPA liability -- including a private right of action letting individuals sue, with statutory damages of $500 to $1,500 per violation.
Sources: FCC.gov, Feb 8, 2024; FCC 24-17, full text (PDF); Mayer Brown analysis.
What actually triggered the ruling
Two days before the January 23, 2024 New Hampshire Democratic primary, robocalls using an AI-cloned voice of President Biden went out telling voters not to vote in the primary and to "save" their vote for November. Political consultant Steve Kramer admitted to creating the scheme, without his client's knowledge. A coalition of 27 state attorneys general, including California's Rob Bonta, had formally petitioned the FCC on January 16, 2024 -- before the incident became national news -- urging exactly the classification the FCC then issued.
The consequences followed later that year. The FCC adopted a $6 million proposed fine against Kramer on September 26, 2024; he has publicly stated he does not intend to pay it, and separately faced New Hampshire state criminal charges. The telecom carrier that transmitted the calls, Lingo Telecom, settled with the FCC for $1 million around August 2024, alongside a mandatory compliance program tied to STIR/SHAKEN caller-ID authentication standards -- a signal that carriers, not just the callers who originate AI robocalls, carry real regulatory exposure.
Sources: California DOJ, Jan 16, 2024; NPR, May 23, 2024; FCC, $6M fine notice; Perkins Coie, Lingo Telecom fine; The Hill.
What this actually requires of businesses using AI voice calling
For any business making outbound calls with an AI-generated voice, the compliance baseline is the same TCPA consent regime that applied to traditional robocalls: prior express written consent for AI-voice calls that are telemarketing or advertising in nature, and prior express consent (non-written) for purely informational calls. A separate FCC rule finalized February 15, 2024 and effective April 11, 2025 also requires businesses to honor consent revocation through any reasonable method -- a caller texting "stop," "revoke" or "cancel" must be honored within a 10-day window, with only one confirmation message allowed.
Beyond that baseline, the picture is still moving. The FCC issued a Notice of Proposed Rulemaking around August 2024, with comments due October 10, 2024, that would require callers to disclose AI voice use at the start of each call and disclose at consent-collection that consent may cover AI-generated calls (with a proposed carve-out for AI used for speech/hearing accessibility) -- but as of the most recent tracking reviewed for this report, that rule had not been finalized into a final requirement, so it should be treated as pending rather than in force. Separately, an unrelated FCC "one-to-one consent" rule for lead generators was vacated by the 11th Circuit on January 24, 2025, which affects how consent chains work more broadly across the TCPA landscape, not just for AI-specific calls.
Sources: NCLC Digital Library, 2024-2025 TCPA developments; Wiley Law, NPRM comment deadlines.
Frequently asked questions
Does the TCPA cover AI-generated voices?
Yes. The FCC's Declaratory Ruling FCC 24-17, adopted February 8, 2024, confirmed that the TCPA's existing ban on calls using an "artificial or prerecorded voice" already covers AI-generated and AI-cloned voices. This was an interpretation of existing law, not a brand-new rule.
What triggered the FCC's ruling on AI voices?
Two days before the January 23, 2024 New Hampshire Democratic primary, robocalls using an AI-cloned voice of President Biden told voters not to vote. A coalition of 27 state attorneys general petitioned the FCC on January 16, 2024 to classify AI voices as "artificial" under the TCPA, and the FCC's ruling followed shortly after.
What penalties has the FCC actually issued for AI robocalls?
The FCC adopted a $6 million proposed fine against political consultant Steve Kramer on September 26, 2024 for the New Hampshire deepfake robocalls. The carrier that transmitted the calls, Lingo Telecom, separately settled with the FCC for $1 million plus a mandatory caller-ID authentication compliance program.
Do businesses have to disclose when a caller is an AI, under current US law?
Not yet, as a hard federal rule. The FCC proposed a rule in mid-2024 requiring callers to disclose AI voice use at the start of each call, but as of the most recent tracking available, that rule had not been finalized. What is already required is standard TCPA consent: prior express written consent for AI-voice telemarketing calls, and prior express consent for informational ones.
Kamaljeet Singh Sidhu
Founder & CEO of Botnira and CEO of The DigiSparrow. Reviews the independent industry research published on this hub.
Read full bio →Related reading
The EU AI Act and Voice AI
What changed for emotion recognition in call centers under the EU's AI Act.
Read the report →AI Voice Cloning Fraud: The Real Numbers
Sourced data on the scale of deepfake voice fraud, and the incidents that actually happened.
Read the report →What Consumers Actually Think of AI Customer Service
Trust, frustration and disclosure preferences from Metrigy, PwC, Twilio and more.
Read the report →