Back to Insights
Article14 min read·January 27, 2026·Updated November 9, 2026

AI Call and Text Compliance 2026: TCPA, FCC and State Laws

AI Call and Text Compliance 2026: TCPA, FCC and State Laws

Last reviewed November 9, 2026. This guide is general information, not legal advice. TCPA law is moving quickly, and courts are splitting on key questions. Check anything you plan to rely on with counsel.

What compliance rules apply to automated outbound AI agent calls?

Outbound AI voice calls count as artificial-voice calls under the TCPA. They need prior express consent, and written consent if they're telemarketing to a wireless number. They must identify the business, offer an opt-out, and follow federal calling hours, Do Not Call rules and state law. AI texts have their own overlapping rules, covered below.

The FCC has already gone after AI voice abuse. In 2024 it fined Steve Kramer $6 million over spoofed robocalls using a deepfake of President Biden's voice. In 2021 it imposed a $225 million forfeiture on Rising Eagle for spoofed robocalls. Those were bad actors. For a legitimate sales team, the likelier risk is a private lawsuit at $500 to $1,500 per call or text.

Regulation runs in three layers: federal, state and carrier. Here's where each stands.

1. Federal: TCPA and FCC rules

The Telephone Consumer Protection Act (TCPA) is the main federal law on automated calls and texts. The FCC writes the rules. Since McLaughlin Chiropractic v. McKesson in June 2025, federal courts no longer have to follow the FCC's reading of the statute, so some questions are now being decided court by court.

  • AI voices are artificial voices. In a February 2024 ruling, the FCC confirmed the TCPA's limits on artificial or prerecorded voices cover AI that generates human-sounding voices. Those calls need prior express consent, identification of the business, and an opt-out method when they're telemarketing.

  • Calling hours. 47 CFR 64.1200(c)(1) bars telephone solicitations before 8 a.m. or after 9 p.m. in the recipient's local time. Some states are stricter.

  • One-to-one consent (vacated). The FCC's rule requiring separate consent for each seller never took effect. The Eleventh Circuit vacated it on January 24, 2025, in Insurance Marketing Coalition v. FCC, and the FCC later deleted the language.

  • Consent revocation (revised, effective November 25, 2026). See the next section.

  • AI disclosure (proposed only). The FCC's 2024 proposal would require callers to disclose AI-generated calls and get consent that specifically covers them. We found no final rule as of November 2026, and the FCC's regulatory agenda lists it under long-term actions.

2. State laws

States add their own telemarketing and AI rules. Texas, Virginia and Florida have the telemarketing rules most likely to bite. Maine, Utah and California have the AI disclosure rules most likely to reach a sales agent. Details below.

3. Carrier requirements

Carriers like AT&T, T-Mobile and Verizon enforce their own rules on top of the law:

  • A2P 10DLC registration. Businesses texting over 10-digit numbers register their brand and campaigns through a messaging provider with The Campaign Registry. Unregistered traffic gets stopped. Twilio, for example, has blocked all unregistered U.S. 10DLC traffic since September 2023. Our A2P 10DLC guide goes deeper.

  • Campaign vetting. Use-case descriptions and sample messages need to match what you actually send. Generic descriptions like "customer updates" are a common reason for rejection.

  • Throughput limits. Registration and vetting affect how many messages a campaign can send.

  • Content filtering and penalties. Carriers filter traffic, and T-Mobile's Code of Conduct lists suspension and restricted quotas for violations. CTIA's Messaging Principles and Best Practices are voluntary guidelines, but carriers expect them to be followed.

What makes an AI texting vendor carrier-approved?

No carrier hands out a badge for AI vendors. Approval attaches to the brand and campaign, and carriers make the call. The Campaign Registry says brands can't register directly and must work with a messaging provider. A vendor that says it's "carrier-approved" should be able to show you whose brand holds the registration, what use case it covers, and how sample messages match what its AI actually sends. A2P 10DLC is a registration, not a certification.

The FCC delayed the original "revoke all" rule twice, most recently to January 31, 2027. Then on September 30, 2026, it adopted FCC 26-67, which replaces the rule. The final text takes effect November 25, 2026. Under the amended 47 CFR 64.1200(a)(10) to (11):

  • Telemarketing opt-outs still cover everything from that caller. A revocation made in response to a telemarketing call or text revokes consent to all future telemarketing from that caller, across calls and texts.

  • Informational opt-outs can be narrow. A caller may treat an opt-out from an informational call or text as applying only to that category.

  • You can designate an exclusive opt-out method. The options are an automated voice or key-press opt-out on a call, a reply to a text using one of seven standardized words (stop, quit, end, revoke, opt out, cancel, unsubscribe), or a website or phone number you designate. You must disclose it clearly on the call or in the text. A text sender using reply-text as its exclusive method must still honor all seven words.

  • If you don't designate a method, you must honor requests made by any reasonable means.

  • The deadline stays at 10 business days. Requests must be honored within a reasonable time, not to exceed 10 business days.

The FCC also opened a Further Notice, with comments due November 25, 2026 and replies due December 28, 2026. It asks about shortening the deadline to 7 business days, how revocations apply across affiliates, and whether callers must offer a way to revoke everything at once. Expect more changes.

The one-to-one consent rule was meant to be the biggest change to telemarketing consent in over a decade. It never took effect. The Eleventh Circuit held it exceeded the FCC's authority under the TCPA. The problem it targeted still drives lawsuits.

Before: the shared consent loophole

A consumer who agreed to marketing from "partners" or "selected businesses" effectively gave consent to dozens or hundreds of companies. Lead generators could bundle that consent and sell the contact to multiple businesses, and those businesses could claim consent even when the consumer had never heard of them.

What the vacated rule would have required

  • Each seller needed separate, explicit consent from the consumer

  • A consent could cover only one seller at a time

  • Calls and texts had to be logically and topically related to the website where consent was given

  • Lead generators could not bundle consent across multiple businesses

Impact on sales teams

The rule is gone, but vague shared consent is still hard to defend. If you rely on purchased or partner leads:

  • Revisit your opt-in flows. Make sure each form clearly names your business.

  • Audit your consent records. Know where each consent came from and what the consumer agreed to.

  • Update your CRM. Store the source, timestamp, and business named for each consent record.

  • Only send leads you can defend to your AI agent. Your team decides which contacts are eligible before the agent ever sees them.

AI-specific compliance requirements

How AI agents handle compliance disclosures in calls

Federal rules don't yet require a call to say "this is AI." They do require an artificial-voice call to state clearly who is responsible for it at the start. Telemarketing calls using an artificial or prerecorded voice must also offer an automated opt-out. Some state laws go further, so disclosing AI up front is the safer default everywhere. In practice that means the agent:

  • Says who it's calling for and that it's an AI, before any sales pitch

  • Makes that disclosure easy to hear, not buried in rapid legal language

  • Gives a recording notice where one is required

  • Tells the person how to opt out on telemarketing calls

  • Never claims to be human, even if pushed

For more on how the law separates AI agents from robocalls, see AI voice agents vs. robocalls.

Does the TCPA apply to AI texts?

It depends on the provision, and courts are split. The FCC treats texts as calls, but under Facebook v. Duguid the TCPA's autodialer limits apply only to systems that use a random or sequential number generator, so a conversational AI text isn't automatically covered there. The Do Not Call rules are less settled. In July 2026, the Seventh Circuit held in Steidinger v. Blackstone Medical Services that texts aren't "telephone calls" under the private right of action in Section 227(c)(5). Other federal courts have ruled the other way, and the Ninth Circuit in Howard v. Republican National Committee treated a text as a call under Section 227(b). Steidinger leaves FCC enforcement and state law intact. Don't plan around the split. Get written consent for marketing texts and honor opt-outs.

Call recording and consent

Many AI agents record conversations. Several states, including California, Florida, Illinois, Maryland, Massachusetts, Montana, New Hampshire, Pennsylvania and Washington, generally require all parties to consent. Others require one party. If a call may cross state lines, treat it as all-party and give notice at the start.

AI guardrails

  • Time-of-day restrictions. Stay within 8 a.m. to 9 p.m. in the recipient's local time, or narrower where state law or your own policy requires.

  • Opt-out handling. The agent should recognize a request to stop and honor it right away, well inside the 10-business-day limit.

  • Message consistency. AI-generated messages should stay consistent with the campaign you registered.

  • Do Not Call scrubbing happens before the agent gets a number. The national registry applies to telephone solicitations, and your team needs to scrub lists before loading them.

State law compliance: telemarketing and AI disclosure

These are the state rules most likely to affect AI calls and texts as of November 2026. It's not a full 50-state survey.

StateLawWhat it doesStatus
TexasSB 140Extends telemarketing registration to textsIn force since Sept 1, 2025. Consent-based texts likely exempt from registration, but not settled
VirginiaVa. Code 59.1-514 (SB 1339)Text STOP or UNSUBSCRIBE must be honored for at least 10 yearsIn force since Jan 1, 2026
FloridaFla. Stat. 501.059Written consent for unsolicited telephonic sales calls using automated dialing or recorded messages. 15 days to stop texting after STOPIn force. No AI-specific consent rule found
Maine10 M.R.S. 1500-DDA chatbot may not mislead a reasonable consumer into thinking it's human unless clearly told otherwiseIn force
UtahSB 226Disclose AI when a consumer clearly asks. Regulated occupations disclose first in high-risk interactionsIn force, with repeal set for July 1, 2027 unless extended
CaliforniaAB 2905 (Pub. Util. Code 2874)Automated call announcements must say if a prerecorded message uses an artificial voiceIn force since Jan 1, 2025
CaliforniaAB 1609Customer service chatbots at businesses over $500 million in revenue can't pose as human, and must offer a path to a personTakes effect Jan 1, 2027
ColoradoSB 26-189Replaced the 2024 AI Act with a narrower law on automated decisionsTakes effect Jan 1, 2027

Texas: SB 140 and text messages

SB 140 changed Texas's definition of "telephone solicitation" to cover texts and other transmissions, and added private lawsuits under the Deceptive Trade Practices Act for some violations. In late 2025 the state took the position in court that consent-based text programs don't need to register, and the Secretary of State's FAQ says the same. K&L Gates notes that's not binding precedent, and no court has ruled on it in a contested case. Cold texts remain squarely covered.

Virginia: 10-year opt-out window

Virginia's law covers telephone solicitations, including texts. A person can opt out by replying STOP or UNSUBSCRIBE, and the request must be honored for at least 10 years. Damages are $500 for a first violation, $1,000 for a second and $5,000 for each one after that, and a court can raise the amount to $5,000 for a willful first or second violation.

California: disclosure and automated decisions

Beyond AB 2905 and AB 1609, California's CCPA automated decisionmaking rules took effect January 1, 2026, with compliance required from January 1, 2027. They're aimed at significant decisions like credit or housing, so a routine sales conversation usually falls outside them. California's 2019 bot disclosure law is written for online communication, so its reach over calls and texts is unclear. Ask counsel if you run high-volume agents into California.

Florida: written consent for automated sales calls

Florida's law requires prior express written consent for unsolicited telephonic sales calls that use an automated system to select and dial numbers or play a recorded message. It doesn't single out AI, and a 2026 bill to amend it (SB 134) died in committee. An AI dialer can still fall under the statute depending on how it selects and dials numbers.

One federal note. A December 2025 executive order aims to challenge some state AI laws, but an executive order doesn't repeal state statutes. Keep complying unless a court says otherwise.

Building a compliant AI sales communication system

Layer 1: Consent and lists (your team owns this)

  • Capture clear consent that names your business and covers calls and texts by AI

  • Record the source and timestamp of every consent

  • Keep consent records well past the period claims can be brought, generally four years for TCPA claims

  • Scrub against the national and state Do Not Call lists before loading contacts

  • Keep your internal do-not-call list for at least 5 years federally, and 10 years for Virginia text opt-outs

Layer 2: AI agent configuration

  • Open every call with identity and AI status, for example "Hi, this is an AI assistant calling for [business]."

  • Set time-of-day restrictions based on the recipient's timezone

  • Configure guardrails that stop messages that break your rules

  • Keep messages consistent with your registered campaign samples

  • Build opt-out handling that stops future communication

  • Decide how your opt-out works under the November 2026 rule, including whether to designate an exclusive method

Layer 3: Carrier registration

  • Register your brand and campaigns through your messaging provider

  • Register specific use cases with detailed descriptions

  • Submit message samples that reflect what AI agents will send

  • Monitor deliverability and adjust campaigns if filtering increases

Layer 4: Documentation and auditing

  • Consent records with timestamps, sources, and the business named

  • AI agent configuration and guardrail settings

  • Carrier registrations and approval confirmations

  • Audit logs of compliance-related actions and decisions

Compliance checklist for AI sales teams

  • ☐ Review all opt-in flows so consent clearly names your business

  • ☐ Store consent source, timestamp and business named for every contact

  • ☐ Configure AI agents to identify the business and disclose AI at the start of calls

  • ☐ Honor opt-outs within 10 business days at the latest, and set retention to at least 10 years where Virginia applies

  • ☐ Confirm telemarketing opt-outs carry across calls and texts

  • ☐ Decide by November 25, 2026 whether to designate an exclusive opt-out method

  • ☐ Complete A2P 10DLC registration through your messaging provider

  • ☐ Register specific use cases with detailed descriptions

  • ☐ Scrub contact lists against national and state Do Not Call lists

  • ☐ Set time-of-day restrictions based on recipient timezone

  • ☐ Review state-specific requirements (TX, VA, FL, ME, UT, CA, CO)

  • ☐ Train your sales team on compliance requirements and AI disclosures

  • ☐ Set a recurring audit, and have counsel review changes

How Apten handles AI compliance

Compliance is shared. Apten gives you controls for what happens in the conversation, and your team owns the rest:

  • Automatic opt-out handling. STOP keywords plus AI detection of opt-out requests, honored per channel across text, call, and email

  • Quiet hours. Configurable outreach hours applied in each lead's local timezone

  • Webhook integration. Webhooks to help sync opt-outs to your system

  • Customizable guardrails. A guardrail system you configure to detect and stop bad messages

  • 10DLC registration assistance. Hands-on help registering your brand and campaigns with carriers

  • Security. SOC 2 Type II, with encryption at rest and in transit

Collecting consent and scrubbing against Do Not Call lists remain your responsibility. Apten honors the opt-outs and contact rules you set. Since telemarketing opt-outs now cover calls and texts from the same caller, check with counsel how your opt-out settings should work across channels.

Tools like Apten help with the operational rules. For how outbound voice works in practice, read our outbound voice AI guide.

Book a demo to see how Apten's opt-out handling, quiet hours and guardrails work.

Sources

Frequently Asked Questions

What compliance rules apply to automated outbound AI agent calls?

The FCC treats AI-generated voices as artificial voices under the TCPA, so outbound AI calls need prior express consent, and telemarketing calls to wireless numbers need prior express written consent. The call must identify the business at the start, offer an opt-out if it's telemarketing, and stay within 8 a.m. to 9 p.m. in the recipient's local time. Do Not Call rules and state telemarketing laws apply on top. This is general information, not legal advice.

Does the FCC require AI disclosure on sales calls?

Not yet as an AI-specific rule. The FCC proposed an AI-generated call disclosure in 2024, but as of November 2026 it has not been finalized. Existing rules already require artificial-voice calls to identify the business, and some states go further. Maine requires disclosure when a chatbot could mislead a reasonable consumer into thinking it's human. Disclosing AI at the start of every call is the safest default.

How do AI agents handle compliance disclosures in calls?

A well-configured AI agent opens by naming the business it's calling for and saying it's an AI, before any sales content. It gives a recording notice where state law calls for one, explains how to opt out on telemarketing calls, and never claims to be human if asked. Guardrails should check each response against those rules and stop anything that breaks them.

What makes an AI texting vendor carrier-approved?

Carriers don't approve vendors as such. They approve the brand and campaign behind the messages, which get registered through a messaging provider in The Campaign Registry. A good vendor helps you register accurate A2P 10DLC campaigns and keeps live messages consistent with the use case you registered. Ask any vendor who holds the registration, how approval works, and what happens if a campaign is suspended.

What happened to the FCC one-to-one consent rule?

It never took effect. The Eleventh Circuit vacated it on January 24, 2025, in Insurance Marketing Coalition v. FCC, and the FCC later deleted the language from its rules. Consent can still legally cover more than one seller, but vague shared consent from lead generators is hard to defend in court.

When does the FCC's revised opt-out rule take effect?

November 25, 2026. An opt-out from a telemarketing call or text revokes consent to all future telemarketing from that caller. An opt-out from an informational message can be limited to that category. Callers can designate an exclusive opt-out method if they disclose it clearly, and must honor requests within 10 business days. The FCC is still taking comment on further changes.

How long do I have to honor an opt-out?

Federal rules require honoring revocations within a reasonable time, no more than 10 business days, and keeping do-not-call requests for 5 years. Virginia requires honoring a text STOP or UNSUBSCRIBE reply for at least 10 years. Florida gives a 15-day window to stop texting after a STOP reply. Many teams honor opt-outs indefinitely so they don't have to track the differences.

What are the penalties for TCPA violations?

The TCPA allows $500 per call or text in statutory damages, and up to $1,500 per violation if the court finds it willful. Class actions multiply that quickly. State laws add their own exposure. Virginia's statute, for example, sets damages of $500 for a first violation, $1,000 for a second and $5,000 for each one after that.

Who is responsible for TCPA compliance when a vendor's AI agent makes the call?

Usually the business the call is made for, even when a vendor places it. Your team owns consent capture and Do Not Call scrubbing. An AI platform can enforce the rules you set during the conversation, like quiet hours, opt-outs and guardrails. Ask counsel how your contracts divide the rest.

A

Apten Team

AI & Sales Automation Experts

Ready to transform your student engagement?

See how Apten's AI-powered SMS and voice agents can help you convert more students with personalized, omni-channel conversations.