AI and compliance

Can AI text and call your leads? The US rules in plain English

What the TCPA, the FCC, state AI laws and CAN-SPAM mean when AI replies to your leads by text, email or phone, plus a checklist to go over with your lawyer.

A new lead fills out your form at 7 p.m. Your AI replies by text in a minute, answers a question and offers a time to talk. That's the dream. But is it allowed?

Mostly, yes, if you set it up the right way. This guide covers the main US rules for AI texts, calls and emails to leads. It ends with a checklist to take to your lawyer.

The TCPA: get consent first

The Telephone Consumer Protection Act (TCPA) is a 1991 federal law. It covers calls made with an "autodialer" (a system that dials numbers automatically) or an artificial or prerecorded voice. The FCC's rules cover texts sent with an autodialer too, as its 2026 order notes.

Marketing calls and texts sent this way face the highest bar. Under the FCC's TCPA rules, they need "prior express written consent." In plain words, that's a signed agreement before you contact them. A valid electronic signature can count.

The agreement must cover your business and the phone number. It must clearly say the person agrees to marketing calls or texts sent with automated tech or an artificial voice. It must also say they don't have to agree in order to buy.

Is your texting tool an autodialer? In 2021 the Supreme Court read that term narrowly. But state laws use their own wording. Florida's law covers sales texts sent with an "automated system" for selecting and dialing numbers. The simple path is written consent at the form.

Example consent line (have your lawyer adjust it): "By checking this box, I agree that [Business name] may contact me about its services at the number above, including automated texts and calls that may use an AI or prerecorded voice. Agreeing is not a condition of purchase. Reply STOP to opt out."

AI voices count as artificial voices

In February 2024, the FCC issued a ruling that AI-generated voices are "artificial" voices under the TCPA. So an AI phone call needs the person's consent first. If the call markets something, it needs written consent.

The FCC rules add must-dos for artificial voice calls. Say who is calling at the start and give a callback number. For sales calls, offer a voice or keypress opt-out within two seconds of saying who you are.

What happened to the "one-to-one" rule

In 2023, the FCC said consent had to name one seller at a time. In January 2025, the federal appeals court for the 11th Circuit struck that rule down before it took effect. The FCC removed it from its rules in August 2025. For your own form, little changes. Still name your business in the consent line.

Opt-outs: honor "stop" fast

Consent isn't forever. Since April 11, 2025, the FCC rules let people take back consent using "any reasonable method." Replies like "stop," "quit," "end," "revoke," "opt out," "cancel" or "unsubscribe" always count. So do other words a reasonable person would read as "stop."

You must honor the request within a reasonable time, and no later than ten business days. You may send one short text confirming the opt-out, with no sales pitch. If it goes out within five minutes, it's presumed fine.

More changes are coming. On September 30, 2026, the FCC adopted an order that lets you name an official opt-out method, like replying STOP. You must state that method clearly on each call or text. An opt-out from marketing still stops all your marketing messages.

That order takes effect 30 days after it's published in the Federal Register. As of mid-October 2026, we couldn't find it there. The FCC is also asking whether to cut the ten-day window to seven. Check the current status with your lawyer.

Teach your AI to spot "stop," "not interested" and "please don't text me." Then it should stop and alert you.

Quiet hours: when the AI may reach out

With some exceptions, the federal rule bars sales calls before 8 a.m. or after 9 p.m. in the called person's local time. Some states are stricter.

  • Florida: its telemarketing law allows sales calls only from 8 a.m. to 8 p.m. in the lead's time zone. It also caps calls at three per 24 hours on the same subject. The law has exemptions, so ask whether it covers you.
  • California: it limits machines that dial and play a recorded message. These calls are banned from 9 p.m. to 9 a.m. California time.

A safe habit: keep AI calls and outbound texts between 8 a.m. and 8 p.m. in the lead's time zone. What about a lead who texts you at 10 p.m.? Ask your lawyer whether an instant reply is fine or should wait until morning.

Register your business texting number

US carriers have their own rules for business texts, called A2P 10DLC. That stands for "application to person" texts sent from a regular 10-digit local number.

As Twilio's guide explains, carriers set it up so business texts are verified and consented to. You register a "brand," which is who you are. Then you register a "campaign," which is what you send and how people opt in and out. With Twilio, unregistered texts get more filtering and extra carrier fees.

If you text from your CRM (one place for every lead, conversation and next step), ask its texting provider how it handles registration. Toll-free numbers have their own approval process.

Do you have to say it's AI?

There's no single federal rule yet for AI texts or chat. In 2024, the FCC proposed that AI-voice calls say so at the start. As of October 2026, we found no final rule. Several states already have rules.

  • California's bot law: sections 17940 to 17943 of the Business and Professions Code ban using a bot online to mislead someone in California about being a bot to push a sale. A clear statement that it's a bot avoids this. "Online" means public websites, apps and social networks.
  • California AI-voice calls: for automated calls that play a recorded message, a live person must first introduce the call. Since 2025, that person must also say if the message uses an AI voice. Ask your lawyer whether a two-way AI voice agent counts.
  • Utah: under its 2025 law, a business using AI to talk with consumers must say so if a person clearly asks. Some licensed professions must disclose up front in sensitive talks. Saying it's AI at the start and throughout gives you a safe harbor.
  • Maine: a 2025 law bars using an AI chatbot, by text or voice, in a way that could make a reasonable consumer think it's a human. You must clearly tell them it isn't.

The easy answer for every state: disclose at the start of every conversation, in every channel.

Example disclosure: "Hi Sam, this is the AI assistant for [Your name] Coaching. I can answer questions and book a call. Want to talk to [Your name] directly? Just say so."

Recording AI calls

Many AI phone tools record calls. Federal law lets one person on the call consent. But some states, like California, require everyone's consent to record a confidential call. The simple fix is to announce the recording at the start of every call.

Email: the CAN-SPAM basics

Email follows a different law, CAN-SPAM. The FTC's compliance guide lists the basics for marketing emails:

  • Use honest "from" details and a subject line that matches the message.
  • Include your valid physical postal address.
  • Give a clear way to opt out, and honor it within ten business days.

Business-to-business email isn't exempt. You stay responsible even if a tool sends it for you.

Guardrails matter as much as the law

Rules on paper don't help if the AI ignores them. It should hand sensitive or unusual questions to you, like complaints, health or money worries, or anything legal. You should be able to pause it any time. See how these guardrails fit into a full AI sales system.

Put your rules in writing, too. For an example, see the AI and Data Use Statement that Nucleus Growth Systems publishes.

A checklist to review with your lawyer

  1. Add an unchecked consent box to every form that asks for a phone number. Name your business and mention automated texts and AI-voice calls.
  2. Save proof of consent: the wording shown, the date, time, form page and phone number.
  3. Have the AI say it's an AI assistant at the start of every chat, text thread and call.
  4. Make sure "stop" and similar replies pause all messages at once and alert you.
  5. Set quiet hours of 8 a.m. to 8 p.m. in each lead's time zone, and cap call attempts.
  6. Register your texting number for A2P 10DLC, or verify your toll-free number.
  7. On AI calls, say who's calling, give a callback number, offer a voice or keypress opt-out and announce any recording.
  8. Put a postal address and a working unsubscribe link in every marketing email.
  9. Send sensitive or unusual questions to a human, and keep a pause button handy.
  10. Check the laws of the states where your leads live.
  11. Review this again when the FCC's 2026 opt-out changes take effect.

Want to see where your follow-up stands today? The free Lead Follow-Up Leak Checklist is a good start. Or book a free 30-minute Lead Follow-Up Review.

Sources

  1. Implications of Artificial Intelligence Technologies on Protecting Consumers from Unwanted Robocalls and Robotexts, Declaratory Ruling (FCC 24-17), Federal Communications Commission, 2024.
  2. 47 CFR 64.1200, Delivery restrictions, Electronic Code of Federal Regulations, checked October 2026.
  3. Rules and Regulations Implementing the Telephone Consumer Protection Act of 1991, Report and Order and Further Notice of Proposed Rulemaking (FCC 26-67), Federal Communications Commission, 2026.
  4. Insurance Marketing Coalition Ltd. v. FCC, No. 24-10277, US Court of Appeals for the Eleventh Circuit, 2025.
  5. Delete, Delete, Delete; Targeting and Eliminating Unlawful Text Messages (final rule), Federal Communications Commission, Federal Register, 2025.
  6. Facebook, Inc. v. Duguid, No. 19-511, Supreme Court of the United States, 2021.
  7. Florida Statutes section 501.059, Telephone solicitation, The Florida Legislature, 2025.
  8. Florida Statutes section 501.616, Unlawful acts and practices, The Florida Legislature, 2025.
  9. California Public Utilities Code section 2872, California Legislative Information.
  10. Programmable Messaging and A2P 10DLC, Twilio Docs, checked October 2026.
  11. Implications of Artificial Intelligence Technologies on Protecting Consumers from Unwanted Robocalls and Robotexts, Notice of Proposed Rulemaking and Notice of Inquiry (FCC 24-84), Federal Communications Commission, 2024.
  12. California Business and Professions Code sections 17940 to 17943, Bots, California Legislative Information.
  13. California Public Utilities Code section 2874, as amended by AB 2905, California Legislative Information, 2024.
  14. S.B. 226, Artificial Intelligence Consumer Protection Amendments (enrolled copy), Utah State Legislature, 2025.
  15. Maine Revised Statutes Title 10, Chapter 241, Communications with Consumers via Artificial Intelligence, Maine Legislature, checked October 2026.
  16. 18 U.S.C. 2511, Interception and disclosure of wire, oral, or electronic communications prohibited, US Government Publishing Office.
  17. California Penal Code section 632, California Legislative Information.
  18. CAN-SPAM Act: A Compliance Guide for Business, Federal Trade Commission.

See what the system would take off your plate, in 30 minutes.

Free Lead Follow-Up Review. No pitch deck. If we're not the right fit, we'll say so.

Book a Free Lead Follow-Up Review