AI Voice Calls and the TCPA: What Small Businesses Must Know

Outbound AI voice calls generally require consent. In February 2024 the FCC ruled that AI-generated voices count as "artificial" voices under the Telephone Consumer Protection Act (TCPA), so a business that calls someone with one needs that person's prior express consent first (Federal Communications Commission, 2024; Federal Communications Commission, 2024). The TCPA sets $500 in damages per violation, and a court can raise that to $1,500 if the violation was willful or knowing (U.S. Code via Cornell LII, 1991). An AI that answers calls your customers make sits on the other side of that line.

This is general information, not legal advice.

What did the FCC decide about AI voices in 2024?

The FCC's February 2024 Declaratory Ruling confirmed that the TCPA's limits on an "artificial or prerecorded voice" cover current AI technologies that generate human voices (Federal Communications Commission, 2024). The reasoning was short: these are artificial voice messages "because a person is not speaking them" (Federal Communications Commission, 2024). The ruling took effect immediately (Federal Communications Commission, 2024), so callers need prior express consent before an AI-voice call, absent an emergency or an exemption (Federal Communications Commission, 2024).

Two details matter for a small business. First, if the call markets anything, FCC rules require prior express written consent: a signed agreement that authorizes your business to send marketing calls to that specific number (Federal Communications Commission, 2024; Electronic Code of Federal Regulations, 2026). Second, the FCC closed the obvious loophole. It said the TCPA "does not allow for any carve out of technologies that purport to provide the equivalent of a live agent" (Federal Communications Commission, 2024). An AI that sounds human and talks back in real time is still an artificial voice.

Every artificial-voice call must also state the identity of your business at the beginning (Electronic Code of Federal Regulations, 2026). Marketing calls must offer an automated opt-out, by voice or key press, within two seconds of that identification (Electronic Code of Federal Regulations, 2026).

Does an AI receptionist answering your phone fall under the same rules?

Generally no, as far as the TCPA's robocall consent rules go. The statute restricts calls a business makes or initiates using an artificial voice (U.S. Code via Cornell LII, 1991), and the FCC's 2024 ruling applied its requirements to "any AI technology that initiates any outbound telephone call" (Federal Communications Commission, 2024). When a customer dials your number and an AI answers, the customer started the call. The line flips the moment your AI dials out.

That includes callbacks to missed calls, appointment reminders, no-show follow-ups, and review requests by phone. Each one is an outbound call that needs consent, and the FCC's consent and opt-out rules cover text messages too (Electronic Code of Federal Regulations, 2026). If you're weighing an AI front desk, our post on AI receptionists covers the inbound side, and A2P 10DLC registration covers business texting.

What has changed since the 2024 ruling?

The one-to-one consent rule is gone. In January 2025 the Eleventh Circuit held that the FCC exceeded its authority with a 2023 rule requiring consumers to consent to one seller at a time, and it vacated that part of the order (U.S. Court of Appeals for the Eleventh Circuit, 2025). The FCC conformed its rules to the decision in an order adopted in July 2025 that took effect August 29, 2025 (Federal Communications Commission (Federal Register), 2025). As of October 2026, written consent is still required for marketing robocalls (Electronic Code of Federal Regulations, 2026).

Three other changes are worth knowing:

• Courts read the statute themselves. In June 2025 the Supreme Court held that in civil enforcement suits, including private TCPA lawsuits like the one before it, district courts are not bound by an agency's reading of a statute and must decide its meaning while giving the agency's view "appropriate respect" (Supreme Court of the United States, 2025). Don't treat that as a loophole. A court would be reading the same word, "artificial," that the FCC read.
• AI disclosure is proposed, not final. In August 2024 the FCC proposed requiring callers using AI-generated voices to disclose, at the beginning of each call, that the call uses AI (Federal Communications Commission, 2024). When we checked the FCC's current rules in October 2026, that requirement had not been added (Electronic Code of Federal Regulations, 2026). It's still a sound habit.
• Opt-outs have a deadline. A customer can revoke consent by any reasonable method, and you must honor it within ten business days at most (Electronic Code of Federal Regulations, 2026).
• Opt-out rules are being revised. On September 30, 2026 the FCC voted to let consumers stop specific categories of robocalls without also stopping informational calls they want, and to let callers designate a clear method for revoking consent (Federal Communications Commission, 2026). Until the new rules take effect, the ten-business-day rule above applies. Our September 2026 brief has the details.

Do you need consent to record AI-handled calls?

It depends on the state. Federal wiretap law (18 U.S.C. 2511) allows recording when one party to the call consents (U.S. Code via Cornell LII, 1968), but some states require everyone's consent. Washington, for example, makes it unlawful to record a private communication by telephone "without first obtaining the consent of all the participants" (Washington State Legislature, 2021). Its statute, last amended in 2021, treats a recorded announcement that the call is being recorded as consent (Washington State Legislature, 2021).

If your AI agent records or transcribes calls, your callers may be in more than one state. Put a short recorded notice at the start of every call, and ask counsel which states' rules apply to you.

What could one AI campaign without consent cost?

The TCPA lets a person sue for actual losses or $500 for each violation, whichever is greater, and a court may raise the award to as much as three times that amount when the violation was willful or knowing (U.S. Code via Cornell LII, 1991). That's up to $1,500 per violation. Because the damages are set by statute rather than by the harm you caused, a small campaign adds up quickly.

Worked example: This is an illustration, not a real business. A home services company loads 300 past customers into an AI voice dialer to promote a fall tune-up. Assumptions: none of the 300 signed a written agreement to receive marketing calls, each number gets one call, and each call counts as one violation. At $500 per violation (U.S. Code via Cornell LII, 1991), exposure is 300 × $500 = $150,000. If a court finds the calls willful or knowing and triples the award (U.S. Code via Cornell LII, 1991), it's 300 × $1,500 = $450,000. A second round of calls the following week would double both numbers.

What to do this week

1. List every automated outbound call and text. Reminders, callbacks, promotions, review requests. Mark which use an AI or prerecorded voice and which are marketing.
2. Pull the consent record behind each list. Marketing calls with an AI voice need a signed written agreement that covers the number you're calling (Electronic Code of Federal Regulations, 2026). No record, no call.
3. Script the first seconds of every AI call. State your business's name at the start (Electronic Code of Federal Regulations, 2026), say the caller is an AI (the FCC has proposed requiring this (Federal Communications Commission, 2024)), and offer an easy opt-out.
4. Make opt-outs stick. Log every "stop" by call or text and honor it within ten business days at most (Electronic Code of Federal Regulations, 2026). Add a recording notice if your agent records or transcribes.
5. Talk to a TCPA attorney before launching outbound AI calls. Bring your call list, your consent language, and your script.

Where Nirvani fits

Nirvani captures incoming calls, texts, social messages, and website chat in one system. Its AI employees answer 24/7 in your business's voice, qualify leads, book appointments into your calendar, follow up on no-shows, and send review requests. Answering is inbound. No-show follow-ups and review requests are outbound, so the consent and opt-out rules above apply to them, and disclosing the AI is good practice. See how our AI agents work and how we handle security.