This article is educational information about United States law for outbound teams, not legal advice. Rules change and facts matter — engage qualified counsel before making compliance decisions.

What NPRM 24-84 proposes

The FCC adopted the Notice of Proposed Rulemaking and Notice of Inquiry (FCC 24-84) on 7 August 2024 and released it on 8 August 2024, in CG Docket No. 23-362. In the Commission’s own words, it proposes “to define AI-generated calls and propose new rules that would require callers disclose to consumers when they receive an AI-generated call.”

A proposed definition of “AI-generated call”

The NPRM proposes defining an AI-generated call as “a call that uses any technology or tool to generate an artificial or prerecorded voice or a text using computational technology or other machine learning, including predictive algorithms, and large language models, to process natural language and produce voice or text content to communicate with a called party over an outbound telephone call.”

Two details are worth noting. The definition is deliberately limited to outbound calls — the FCC states that “the TCPA’s requirements do not extend to technologies used to answer inbound calls,” so inbound AI receptionists sit outside the proposal. And “call” includes text messages, which courts and the Commission have long treated as calls under the TCPA.

Disclosure at consent time and at the start of each call

The NPRM proposes three linked disclosure duties:

• At consent time (voice): “clear and conspicuous disclosure” that consent to receive artificial and prerecorded calls may include consent to receive AI-generated calls.
• At consent time (text): the same duty for autodialed text messages that include AI-generated content.
• On every call: callers using AI-generated voice would be required, “at the beginning of each call, clearly disclose to the called party that the call is using AI-generated technology.” The FCC even sought comment on whether a special tone or badge should accompany that disclosure.

An accessibility exemption

The NPRM also proposes to exempt from the TCPA’s consent and identification requirements artificial or prerecorded voice calls “made by an individual with a speech or hearing disability using any technology, including artificial intelligence technologies, designed to facilitate the ability of such individuals to communicate over the telephone” — so that rules against AI abuse do not deter assistive tools.

The timeline so far — and why it is still not law

The sequence to date:

• 7–8 August 2024: NPRM adopted and released (FCC 24-84, CG Docket 23-362).
• 10 September 2024: published in the Federal Register; per law firm Wiley Rein’s client alert, initial comments were due 10 October 2024 and replies 25 October 2024.
• Late 2024 onwards: the record closed. No Report and Order finalising the proposals has been released since — as of August 2026, none of the proposed AI-disclosure duties is in force.

The regulatory climate also shifted after the NPRM was adopted. Under Chairman Brendan Carr, the Commission opened a broad deregulatory proceeding in March 2025 — the “In Re: Delete, Delete, Delete” docket (GN Docket No. 25-133), which, as law firm Covington & Burling describes it, invited the public to identify FCC rules that should be removed to alleviate “unnecessary regulatory burdens”. A commission focused on deleting rules is not an obvious sponsor for adopting new ones. Plan on the basis that the NPRM could be finalised, modified, or left dormant indefinitely.

One caution: do not confuse this pending NPRM with the FCC’s “one-to-one consent” rule for lead generators, which the Eleventh Circuit vacated in January 2025 — a separate rule that was finalised and then struck down. We cover that vacatur in our AI cold-calling software guide.

What is already binding today, regardless of the NPRM

The February 2024 declaratory ruling: cloned voices are “artificial” voices

In a Declaratory Ruling adopted 2 February 2024 and released 8 February 2024 (also CG Docket 23-362), the FCC confirmed “that the TCPA’s restrictions on the use of ‘artificial or prerecorded voice’ encompass current AI technologies that generate human voices.” The consequence is immediate: calls using AI-generated or cloned voices “require the prior express consent of the called party” absent an emergency purpose or exemption — and prior express written consent for telemarketing. If your programme uses voice cloning, this ruling is the controlling authority today; our guide to voice cloning for business calls goes deeper.

The TCPA identification baseline: 47 CFR § 64.1200(b)

Existing FCC rules (see 47 CFR § 64.1200, hosted by Cornell’s Legal Information Institute) already require every artificial or prerecorded-voice message to:

• state clearly, at the beginning of the message, the identity of the business, individual or other entity responsible for initiating the call;
• state clearly, during or after the message, the telephone number of that business or individual (not the autodialer’s number); and
• for telemarketing calls, provide an automated, interactive voice- and/or key-press-activated opt-out mechanism.

Because the February 2024 ruling folds AI voices into the “artificial voice” category, these identification and opt-out duties apply to AI voice calls right now. The NPRM would add an explicit “this call uses AI” disclosure on top — it would not replace this baseline.

State-level disclosure laws already exist

Federal rules are not the whole picture. Two verified examples:

• California’s B.O.T. Act (Business & Professions Code §§ 17940–17943) makes it unlawful to use a bot to mislead a person online about its artificial identity in order to incentivise a purchase or influence a vote — with a defence where the bot’s non-human status is disclosed clearly and conspicuously. Its scope is online communications rather than telephone calls, but it signals where state law is heading.
• Utah’s Artificial Intelligence Policy Act (SB 149, effective May 2024) required businesses to disclose generative-AI use when consumers asked, with proactive duties for state-licensed occupations. According to law firm Davis Polk’s client update, a 2025 amendment (SB 226, effective 7 May 2025) narrowed this to disclosure upon a clear and unambiguous request plus proactive disclosure in “high-risk” interactions — while adding a safe harbour for businesses that disclose AI use clearly at the outset anyway.

The pattern across jurisdictions is consistent: disclosure obligations are arriving piecemeal, and the safest posture is the one Utah’s safe harbour rewards — say up front that it’s AI. For the broader multi-jurisdiction view (Australia’s Spam Act, the TCPA, GDPR), see our AI outreach compliance overview.

Why voluntary up-front disclosure is becoming standard practice anyway

Even with the federal rule pending, many outbound teams have moved to voluntary AI disclosure at the top of the call, for practical rather than purely legal reasons. The FCC has told you where the puck is going — the NPRM’s stated view is that “consumers have a right to know they will be interacting with AI and to decide whether to continue that” interaction — so building scripts around that expectation now means no scramble if a final rule lands. Disclosure is also a trust mechanism: a prospect who discovers mid-conversation that the “person” is synthetic tends to react worse than one told up front. And state regimes like Utah’s already reward up-front disclosure with an explicit safe harbour.

Where does a platform like Zian AI fit? Zian AI is an autonomous AI sales agents platform running live phone, SMS, email and WhatsApp outreach in 30+ languages, with voice cloning among its capabilities. That last point matters: under the February 2024 ruling, a cloned voice is an “artificial” voice, so US calls made with cloned voices sit squarely inside the TCPA’s consent framework. Zian AI does not make your calling programme compliant — no software vendor can — and disclosure language, consent capture and calling lists remain your team’s responsibility, ideally shaped with counsel. What the platform controls is execution: SmartReach AI™ orchestrates message, channel and timing by country, industry and profile, which is how teams keep US phone scripts (with their identification and disclosure lines) distinct from, say, Australian ones. For the operational comparison with traditional teams, see AI voice agents vs call centres.

A practical operating checklist while the rule is pending

1. Treat every AI or cloned voice call as an “artificial voice” call — prior express consent for informational calls, prior express written consent for telemarketing.
2. Keep consent records that would survive scrutiny: who consented, when, to what wording, through which channel.
3. Script the existing identification baseline into every call: the responsible entity named at the beginning, a working callback number, an automated opt-out on telemarketing calls.
4. Add a voluntary AI disclosure line at the top of the call — plainly stating the caller is an AI assistant acting for your company. This anticipates the NPRM and aligns with state safe-harbour logic.
5. Update consent-capture language now to cover AI-generated voice and text explicitly — the NPRM’s consent-time proposal shows exactly what regulators want to see.
6. Honour opt-outs fast and maintain DNC hygiene across every channel.
7. Check state law for each state you dial into — Utah and California are not the only movers.
8. Assign someone to watch CG Docket 23-362 so a final rule doesn’t surprise you mid-quarter.

Already law vs proposed: the comparison

Requirement	Already law today (US)	Proposed under NPRM 24-84 (not in force)
AI disclosure on the call	No federal duty to announce “this is AI”; some state laws (e.g. Utah) impose disclosure duties in defined situations	Clear disclosure at the beginning of each AI-generated voice call
Consent	Prior express consent for artificial/prerecorded-voice calls — including AI and cloned voices per the Feb 2024 ruling; written consent for telemarketing	Consent-time disclosure that consent may include AI-generated calls; written agreements to state AI content explicitly
Caller identification	Responsible entity named at the beginning of the message; callback number during or after it (47 CFR § 64.1200(b))	Unchanged — the AI disclosure sits on top of the existing duty
AI-generated texts	Autodialed texts require consent under the TCPA; no AI-specific federal disclosure duty	Consent-time disclosure for autodialed texts with AI content (no on-message disclosure proposed)
Accessibility	No AI-specific federal exemption	Exemption from consent and identification duties for people with speech or hearing disabilities using assistive technologies, including AI
Penalties / enforcement basis	TCPA liability — FCC enforcement plus a private right of action with statutory damages per call	Same TCPA framework; new disclosure rules would add grounds for liability

What would make it final — and what to watch

For the proposals to become binding, the FCC would need to adopt a Report and Order in CG Docket 23-362, publish it in the Federal Register and set effective dates — none of which has happened. Signals worth watching: the item appearing on an FCC open-meeting agenda or on circulation; movement in the Delete, Delete, Delete proceeding that touches TCPA rules; renewed congressional interest in AI-robocall legislation; and any high-profile AI voice-fraud incident, since public pressure on this docket has tracked incidents like the AI voice-cloning robocall that impersonated President Biden before the 2024 New Hampshire primary — cited in the Chairwoman’s statement accompanying the NPRM.

FAQ

Is FCC NPRM 24-84 in force?

No. It is a Notice of Proposed Rulemaking — a proposal. The FCC adopted it on 7 August 2024 and released it on 8 August 2024 (the dates appear on the first page of the official document at docs.fcc.gov), comments closed in October 2024, and as of August 2026 no final rule has been adopted. None of its proposed disclosure duties currently binds callers.

Do AI voice calls already require consent in the US?

Yes. The FCC’s February 2024 declaratory ruling confirmed that AI technologies that generate human voices fall within the TCPA’s “artificial or prerecorded voice” restrictions, so such calls require the called party’s prior express consent (written consent for telemarketing) absent an exemption. That applies to cloned voices too.

What must an AI-voice call disclose today, before any new rule?

Under existing FCC rules, every artificial or prerecorded-voice message must state the identity of the responsible business or individual at the beginning of the message and provide that entity’s telephone number during or after it; telemarketing messages must also offer an automated opt-out mechanism. There is no current federal duty to announce that the voice is AI-generated — that is what the NPRM would add.

Does the NPRM cover SMS as well as voice?

Yes, partly. The FCC treats a text message as a “call” under the TCPA, and the NPRM proposes a consent-time disclosure for autodialed texts containing AI-generated content. The beginning-of-call disclosure, however, is proposed for voice calls only.

Is this the FCC rule that a court struck down in January 2025?

No. The Eleventh Circuit vacated the FCC’s separate “one-to-one consent” rule for lead-generated consent in January 2025. That rule had been finalised and was then struck down; NPRM 24-84 is a different, still-pending proposal in a different area — AI disclosure. Our cold-calling software guide covers the vacatur in detail.