The FCC Just Changed What "STOP" Means: What Every AI Reminder, Review-Request and Reactivation Text Must Do Now

What changed, in one paragraph

On 30 September 2026 the FCC adopted a Report and Order in CG Docket 02-278 (FCC 26-67, released 1 October) that rewrites 47 CFR 64.1200(a)(10), the consent-revocation rule. A caller may now treat a revocation sent in reply to an informational message as applying only to that category of informational messages. A revocation sent in reply to anything containing an advertisement or constituting telemarketing still revokes consent to all marketing calls and texts from that sender. Callers may also designate an exclusive way to opt out, provided they disclose it clearly and conspicuously in each message. The new rules take effect 30 days after Federal Register publication and supersede the 31 January 2027 date the old "revoke all" provision had been waived to.

The order says nothing about AI; it applies to any robotext, which is what an AI-sent reminder is. It is also permissive: it lets a sender scope a STOP, it does not require anyone to. This is an operator's explainer, not legal advice. The duties below are what the rule text says; whether a given message of yours is "informational" or "telemarketing" is a question for counsel, and it is the question the whole rule now turns on.

The timeline, so you know which rule you are under

• 8 February 2024. FCC 24-17 (adopted 2 February, released 8 February; CG Docket 23-362) confirms that an AI-generated voice is an "artificial or prerecorded voice" under the TCPA, so an outbound AI call needs prior express consent. This has not changed.
• 11 April 2025. The 2024 revocation rule, 64.1200(a)(10), takes effect: consumers may revoke "by any reasonable means," and revocations must be honoured within a reasonable time not exceeding ten business days.
• 6 January 2026. The Consumer and Governmental Affairs Bureau (DA 26-12) waives the "revoke all" part of that rule until 31 January 2027.
• 9 September 2026. The FCC circulates a draft order replacing "revoke all" with category-specific revocation for informational messages.
• 30 September 2026. The Commission adopts it. The ABA Banking Journal reports the vote as 3-0. Covington's summary describes the final text as "largely unchanged from the draft."
• 1 October 2026. Released as FCC 26-67. Effective 30 days after Federal Register publication; comments on the accompanying Further Notice are due 30 days after publication, replies 30 days after that.

Check the Federal Register for the publication date; that is the date that starts the 30-day clock.

The three buckets

The permission only works if your platform knows which bucket each outbound message is in. Here is how the three SMB text workflows on this site sort, using the rule's own distinction between informational messages and messages containing an advertisement or constituting telemarketing.

The operator's trap is the second row. Many review-request templates end with "and here is 10% off your next visit." That sentence moves the message out of the informational bucket, and a STOP in reply now turns off every marketing text you send that customer, not just review requests.

What your platform needs if you want to use the new rule

Tag every outbound message with its category, if you intend to scope a STOP. The order does not require tagging; it is the precondition for using the permission it grants. Not "SMS" versus "voice"; reminder versus review versus marketing. If your platform cannot carry that tag, it cannot scope a STOP, and you are back to treating every STOP as revoke-all, which remains lawful and remains the safe default.

Disclose the opt-out method in each message, if you rely on an exclusive method. The rule lets you designate one or more of three methods: an automated voice or key-press opt-out on a call; a reply text using one of seven standardised words ("stop," "quit," "end," "revoke," "opt out," "cancel," or "unsubscribe"); or a website or phone number you provide. You must "clearly and conspicuously disclose the designated method to revoke consent on the call or in the text." Covington notes the final order clarifies that disclosing even one of the seven words in each text satisfies that. If you do not designate a method, you must honour revocation by any reasonable means, as before.

Support reply texts, or disclose that you cannot. If your texting protocol does not allow replies, each text must say so and give a reasonable alternative way to revoke. The Further Notice asks whether to require two-way texting outright, so build for it.

Honour within ten business days. The rule keeps "not to exceed ten business days." The Further Notice asks whether to cut that to seven. Automated opt-out processing should already be same-day.

Keep the consent artifact. For informational messages, the record of the transaction and the number the customer gave you. For marketing, prior express written consent that names the number and the sender. The rule does not change what consent you need; it changes what a STOP takes away.

What this means for the AI layer

An AI agent sending or placing these messages is still a robocall or robotext under the TCPA. Two things follow.

First, if the agent speaks, the February 2024 ruling applies: AI voice is artificial voice, and outbound AI calls need prior express consent. Platforms such as Vapi, Retell and Bland supply the voice; your consent record is yours to keep.

Second, the category tag has to be set where the message is composed. If a reactivation campaign in GoHighLevel or a flow in Klaviyo or Postscript sends through Twilio, the system that knows it is a marketing message is the campaign tool, not the carrier. Practice platforms like Weave and Podium sit on both sides of the line, sending reminders and promotions from the same number; ask each vendor to show you how a STOP is scoped and logged, not just that it exists.

The state overlay

Two state rules add disclosure duties on top of the federal consent rules, and they are not about STOP.

• Texas, HB 149, effective 1 January 2026. A health care provider using AI in a patient interaction must make "a clear and conspicuous disclosure" that the patient is interacting with an AI system "no later than the date on which the service or treatment is first provided." Attorney General enforcement; 60-day cure period; no private right of action.
• Utah, SB 226 and related bills, effective 7 May 2025. Generative-AI disclosure is required on "clear and unambiguous request," and proactively in "high-risk" interactions in regulated occupations (collecting sensitive personal data, or giving financial, legal, medical or mental-health advice). The Utah AI Policy Act sunsets 1 July 2027 unless extended.
• California, AB 1609, approved 28 September 2026, operative 1 January 2027. Businesses over $500,000,000 in national gross annual revenue must disclose that a customer-service chatbot is AI and make a good-faith effort to connect a customer to a human within 15 minutes on request, or offer an appointment within one business day; civil penalties are up to $5,000 for an initial violation and up to $10,000 for each subsequent one. Below that line it does not apply to you; above it, it does.

The FCC's own AI-disclosure proceeding (CG Docket 23-362, proposed in 2024) had not produced a final rule as far as we could find on 3 October 2026. Do not wait for it; the state rules are in force now.

What breaks in month three

The STOP list is one list. Your platform stored every opt-out as a global suppression, which was correct under revoke-all and is still lawful now, but it means the customer who stopped promotions also stopped getting reminders and missed an appointment. Category-scoping is a permission, not a duty; use it only if the disclosure and tagging are in place.

The template got a coupon. Someone added an offer to the review-request text to lift response rates. The message changed buckets and nobody told compliance.

The consent record is a checkbox nobody can produce. When a complaint comes, you need the number, the date, the wording the customer agreed to, and the message they replied STOP to. Export that from your platform today and see what comes back.

Next step

Fix the tagging first, then the templates. The new AI disclosure and consent compliance workflow produces the per-channel disclosure script, the consent log and the opt-out routing once, then re-uses them across every agent. The three workflows this rule touches most are appointment reminders, review request automation and the reactivation campaign; missed-call text back and the post-service follow-up sequence carry the same tag. Our 2025 posts on reactivating cold leads by SMS and review generation pre-date this rule; read them with this page open. The free 10 AI Workflows Every SMB Can Steal covers the rest of the stack. Hubs: home services, dental.