AI Phone Answering: What the FCC and the States Say

“Can an AI answer my phone and call people back?” Those are two different questions, and federal law treats them differently. An AI that picks up when a customer calls you is one thing. An AI voice that dials a customer is another, and the Federal Communications Commission ruled in 2024 that it counts as an “artificial” voice under the robocall law. Here is where the lines sit, what the FCC has proposed but not yet adopted, what one state already requires, and what any AI on your line should never say.

Answering: the robocall rules are about outbound calls

The Telephone Consumer Protection Act’s limits on artificial and prerecorded voices apply to calls a business makes. The FCC said so directly in its August 2024 proposal on AI calls (FCC 24-84): the TCPA’s prohibition on using an artificial or prerecorded voice message “extends only to outbound calls that are ‘made’ or ‘initiated’ by the caller,” and “The TCPA’s requirements do not extend to technologies used to answer inbound calls.”

So an AI receptionist that answers when a homeowner rings your business line is not a robocall under that law. That does not make it rule-free. Recording and transcription laws still apply, state disclosure laws may apply, and anything the system tells a caller is still your business talking. Each of those is below.

Calling back: an AI voice is an “artificial” voice

On February 2, 2024, the FCC adopted a declaratory ruling (FCC 24-17, effective on its release on February 8) confirming that the TCPA’s restrictions on the use of “artificial or prerecorded voice” encompass current AI technologies that generate human voices. The consequence, in the ruling’s words: “callers must obtain prior express consent from the called party before making a call that utilizes artificial or prerecorded voice simulated or generated through AI technology.” Voice cloning is covered too, and the FCC added that the TCPA “does not allow for any carve out of technologies that purport to provide the equivalent of a live agent.”

The ruling says these requirements apply to “any AI technology that initiates any outbound telephone call using an artificial or prerecorded voice.” In practice that means:

• Consent first. The statute bars calls using an artificial or prerecorded voice to a cell phone, and to a residential line, without the called party’s prior express consent, absent an emergency or an exemption (47 U.S.C. §227(b)(1)(A) and (B)).
• Written consent for marketing. If the AI call includes or introduces an advertisement or is telemarketing, the FCC’s rules require prior express written consent (47 CFR §64.1200(a)(2) and (3), cited in the ruling).
• Say who is calling. Every artificial or prerecorded voice message must, at the beginning, “state clearly the identity of the business,” using the name registered with the state, and during or after the message state a telephone number for the business (§64.1200(b)(1) and (2)). Telemarketing messages also need an automated opt-out (§64.1200(b)(3)).
• The cost of getting it wrong. A person who receives a call in violation may sue for actual losses or $500 per violation, whichever is greater, and a court may raise that to up to three times the amount for a willful or knowing violation (§227(b)(3)).

Whether a customer who called you has consented to an automated callback depends on the facts, and this article will not guess for you. The clean way is to ask. Say a gutter company’s AI answers an after-hours call: before it hangs up, it asks, “Is it all right if our automated assistant calls you back at this number about your estimate?” and the yes, the number and the time go into the job record. A person on your crew calling back from their own phone is not an artificial voice at all.

Texts follow their own consent rules, covered in TCPA in Plain English and Texting Customers in 2026.

What the FCC has proposed, which is not yet a rule

In August 2024 the FCC proposed going further (FCC 24-84). It proposed defining an “AI-generated call” as one that uses AI to produce voice or text content “to communicate with a called party over an outbound telephone call,” and three disclosures:

• When you collect consent to artificial or prerecorded calls, say clearly that the consent may include AI-generated calls.
• For autodialed texts with AI-generated content, say the same when collecting consent.
• On an AI-voice call, “at the beginning of each call, clearly disclose to the called party that the call is using AI-generated technology.”

A proposal is not a rule. Before you rely on the current position, check the FCC’s docket for CG Docket No. 23-362 to see whether a final order has been adopted. Building the disclosures in now costs nothing and leaves nothing to change if it is.

The states: Utah as the clearest example

State laws on AI are newer and vary, so check yours. Utah has one of the clearest, and its wording is about interacting with consumers, not about who dialed whom, so read it as covering the AI that answers your line too:

• Disclose when asked. A supplier that uses generative AI to interact with someone in a consumer transaction must disclose that the person is interacting with generative AI and not a human, if the person asks or prompts about it with a “clear and unambiguous request” (Utah Code §13-77-103(1)).
• Licensed occupations, high-risk talks. Someone providing services in a “regulated occupation,” one regulated by Utah’s Department of Commerce that needs a state license or certification, must prominently disclose AI use when the interaction is “high-risk,” which the law defines to include collecting sensitive personal information such as financial data. That disclosure is given verbally at the start of a spoken interaction (§13-77-101 and §13-77-103(2) and (3)).
• The safe harbor. No enforcement action lies under §13-77-103 if the AI clearly and conspicuously discloses, at the outset and throughout the interaction, that it is generative AI, is not human, or is an AI assistant (§13-77-104).
• Enforcement. Utah’s Division of Consumer Protection enforces it, with administrative fines of up to $2,500 per violation (§13-77-105).

California’s bot law is narrower but worth knowing if you put a chat assistant on your website. It makes it unlawful to use a bot to communicate with a person in California online with the intent to mislead them about its artificial identity in order to incentivize a purchase or sale, and a person using a bot is not liable under it if they disclose that it is a bot (Cal. Bus. & Prof. Code §17941). “Online” there means a public-facing website or app (§17940), so it speaks to the chat box on your site rather than to your phone line.

The common thread is easy to meet: have the system say, in its first sentence, that it is an automated assistant for your business, and never let it claim to be a person.

Recordings and transcripts

Check whether your AI answering system records or transcribes calls. Recording consent is a state-by-state rule, and some states require every party’s consent. Washington’s statute, for one, requires “the consent of all the participants,” and counts consent as obtained when one party announces to the others that the call is about to be recorded, provided the announcement is recorded too (RCW 9.73.030). Before you switch one on, read Recording Customer Calls: One-Party and All-Party States and put the announcement in the greeting.

What it should say, and what it should never promise

The FTC Act declares “unfair or deceptive acts or practices in or affecting commerce” unlawful (15 U.S.C. §45(a)(1)), and the agency has been plain that software is no excuse. Announcing its AI enforcement sweep in September 2024, the then FTC Chair, Lina M. Khan, said, “Using AI tools to trick, mislead, or defraud people is illegal. The FTC’s enforcement actions make clear that there is no AI exemption from the laws on the books.” Treat every sentence your AI speaks as a sentence you said.

A good script does a short list of things well:

• Names your business and says it is an automated assistant, then repeats that if asked
• Takes the caller’s name, number, address and what they need
• Books only into times you have actually opened
• Hands anything urgent to a person by your written rule, and tells anyone describing danger to get to safety and dial 911
• Offers to take a message for a person whenever the caller wants one

And it never does these:

• Quotes a firm price for work nobody has seen
• Promises an arrival time, a start date or a warranty term you have not set
• Says a licensed or certified tech is coming unless that is true for that job
• Claims to be a person, or invents a name that sounds like one
• Takes a card number by voice; see Taking Card Payments Safely

Then test it the way a customer would. Call it at night, ask whether it is a robot, ask for a price, describe a burst pipe, and read the transcript. If you are still weighing an AI against a staffed service, Answering Services for Contractors has the trial checklist; most of it applies to software too.

The short version

• An AI answering inbound calls is outside the TCPA’s artificial-voice rules, per the FCC
• An AI voice calling out needs the called party’s prior express consent, written consent for marketing, and the business’s name at the start
• The FCC’s 2024 disclosure proposal is not yet a rule; check the docket
• Utah requires disclosure on a clear request, more for licensed occupations in high-risk talks, and gives a safe harbor for saying it up front
• Check recording consent before the first call is recorded
• Whatever the AI says, your business said

General information, not legal or tax advice. Rules change and vary by state; check with the agency named above or a professional for your situation.