AI voice calls fall under the TCPA the same way robocalls do. Since February 2024 the FCC has treated AI-generated voices as an “artificial or prerecorded voice,” so outbound AI calls need the called person’s prior consent, and marketing calls need prior express written consent. Inbound calls answered by an AI agent follow different rules.
This guide explains what the FCC decided, which consent level each type of AI call needs, what an AI call must say, how calling hours and Do Not Call lists apply, why inbound AI is different, and what to check before you switch on outbound Voice AI in GoHighLevel. It is written for US businesses and agencies, and it is general information, not legal advice.
AI voice calls and the TCPA: what did the FCC decide?
In Declaratory Ruling FCC 24-17, adopted February 2, 2024 and released February 8, 2024, the FCC confirmed that the TCPA’s limits on “artificial or prerecorded voice” calls cover voices generated by AI, including cloned voices. The ruling did not create a new law. It clarified that existing TCPA rules already apply.
That clarification matters for every business testing an AI caller. An AI agent that dials a lead, speaks with a synthetic voice and holds a live conversation is still, in the FCC’s view, an artificial-voice call. How natural the voice sounds, or how well the agent handles questions, does not change the category.
The ruling also gives state attorneys general a clearer path to act. They can bring TCPA cases against callers who misuse AI voices, on top of the private lawsuits the TCPA already allows.
What the ruling did not change
The TCPA’s existing exemptions still apply. Calls made for an emergency purpose, and certain exempt calls defined in the FCC’s rules, keep the treatment they had before. The ruling also did not ban AI calling. It put AI calls inside the consent framework that already governed prerecorded calls.
What consent do outbound AI voice calls need?
Outbound AI voice calls need the called party’s prior express consent at minimum. If the call markets a product or service, the FCC’s rules require prior express written consent. The consent belongs to the business making the call, and it has to cover the kind of call you place.
The consent level depends on what the call is for, not on the technology. Here is how the common call types map:
| AI call type | Examples | Consent level under the FCC rules |
|---|---|---|
| Telemarketing | Offers, promotions, sales follow-up, reactivating old leads with a pitch | Prior express written consent |
| Informational | Appointment confirmations, service updates, a call back the lead asked for | Prior express consent (to cell phones) |
| Emergency | Genuine safety or health alerts | Exempt from the consent requirement |
Prior express written consent is a signed agreement, which can be electronic, that clearly authorizes the business to deliver marketing calls using an artificial or prerecorded voice to a specific number. It must not be a condition of purchase. In practice it is captured on a web form with a checkbox and disclosure text next to the phone field, the same way SMS opt-in language is captured for texting.
Does consent from an old form still count?
Only if the wording covered the calls you now plan to make. A form that agreed to “contact about my quote” by text may not cover a marketing call from an AI voice. Before you point Voice AI at an aged list, read the exact consent language each contact agreed to and when.
What happened to the one-to-one consent rule?
The FCC adopted a rule requiring consent to be given to one seller at a time, but the Eleventh Circuit vacated it on January 24, 2025. Shared lead-generation consent still carries real risk, because the consent must clearly name or cover the business that calls.
New to GoHighLevel? Try it free for 30 days
HighLevel's standard free trial is 14 days. Through our partner link you get 30 days plus HighLevel's free onboarding bootcamp. Rather have it built for you? Our US team sets up GoHighLevel accounts, websites, funnels and automations.
Partner link: we may earn a commission at no extra cost to you. Disclosure
What must an AI voice call say?
An artificial-voice call must identify the business responsible for the call at the start and give a phone number during or after the message. Telemarketing calls with an artificial or prerecorded voice must also offer an automated way for the person to opt out during the call.
These identification rules come from the FCC’s existing artificial-voice requirements in 47 CFR 64.1200(b), so they apply to AI agents as well. The practical checklist for an outbound AI script is short:
- Who is calling: the legal or registered business name in the opening line, not only a first name for the agent.
- A callback number: a working number the person can use to reach the business.
- An opt-out: on marketing calls, a clear way to say “stop calling” that the agent honors, plus a record that marks the contact as Do Not Call.
- Honest answers: if the person asks whether they are talking to a human, the agent should not claim to be one.
Do you have to say the caller is an AI?
There is no final federal rule that requires a specific AI disclosure on every call yet. The FCC proposed AI-call disclosure rules in a 2024 rulemaking (FCC 24-84), and that proposal has not become a final rule. Some states have their own AI or chatbot disclosure laws, so many businesses disclose AI at the start of every call anyway. It costs little and removes an argument later.
Do calling hours and Do Not Call lists apply to AI calls?
Yes. AI telemarketing calls follow the same calling-time and Do Not Call rules as any telephone solicitation. Federal rules bar solicitation calls before 8 a.m. or after 9 p.m. at the called person’s location, and numbers on the National Do Not Call Registry need the right consent or an existing business relationship.
Three lists matter for AI calling:
- The National Do Not Call Registry: scrub marketing lists against it unless you hold prior express written consent for that number.
- Your internal Do Not Call list: every “don’t call me” request goes on it, and the FCC’s rules give you 10 business days to process it.
- State lists and hours: several states keep their own do-not-call lists or set tighter calling hours than the federal window. Our guide to TCPA texting rules covers how state rules stack on the federal ones for SMS.
Since April 11, 2025, the FCC’s rules also say a person can revoke consent by any reasonable means, and the request must be honored within 10 business days. If someone tells your AI agent to stop calling, that is a revocation, and it should flow straight into Do Not Disturb for calls.
Is an inbound AI receptionist treated differently?
Yes. The TCPA’s artificial-voice consent rules govern calls you place to people. When a customer calls your business and an AI agent answers, the customer started the call, so outbound consent rules do not apply in the same way. Call recording and disclosure laws still do.
That is why an AI receptionist on GoHighLevel is usually the lower-risk place to start with Voice AI. The agent answers, qualifies and books, and every conversation begins with a caller who chose to call.
What about recording inbound AI calls?
AI calls are usually recorded and transcribed. Some states, including California and Florida, require every party to agree before a call is recorded. A short line in the greeting, such as “this call may be recorded,” is the common way businesses handle it. Confirm the requirement for the states you serve.
When does an inbound flow turn into outbound calling?
As soon as the system dials out. A callback the caller asked for is informational and needs prior express consent. A follow-up sales call days later is telemarketing and needs prior express written consent. The same AI agent can be inside the rules on Monday’s inbound call and outside them on Friday’s follow-up.
How do you use GoHighLevel Voice AI safely?
GoHighLevel Voice AI can place outbound calls, but HighLevel’s help center is clear that the business, not HighLevel, is responsible for consent. The platform adds guardrails such as a calling window and volume caps. Consent records, scripts, Do Not Call handling and state rules remain yours.
Here is what HighLevel’s help center currently describes for outbound Voice AI, and what it means for you:
- Consent is your job: HighLevel’s current articles say the platform does not validate each contact’s consent before an outbound Voice AI call, and that accepting HighLevel’s outbound calling terms is not consent from the contact.
- Calling window: outbound Voice AI calls are scheduled between 8:00 AM and 8:00 PM in the contact’s phone-number time zone, and calls outside it move to the next eligible time. Area codes do not always match where a person lives now, so this window is a guardrail, not a guarantee.
- Volume and frequency caps: HighLevel limits outbound Voice AI calls per location per minute and per day, and how often one number can be called. Check the current limits in the help center before you plan a campaign.
- AI disclosure setting: outbound agents have a configurable AI disclosure in the agent settings. Turn it on and read the greeting before you publish the agent.
- Domestic numbers: outbound calls are limited to same-country numbers, and they go out from your LC Phone or connected number.
The rest of the stack matters too. A number with a good reputation, SHAKEN/STIR attestation and a registered caller name is more likely to be answered and less likely to be labeled spam. Our guide to GoHighLevel verified caller ID covers CNAM, Voice Integrity and how they fit with your numbers.
Keep consent where the agent can see it
Store how and when each contact consented on the contact record: the form, the checkbox wording, the date and the source. Use Do Not Disturb per channel so a call opt-out stops calls without silencing email. If your leads come through webhooks or the API from another system, make sure the consent fields travel with them, which is part of what our GoHighLevel API and webhook integration service sets up.
Texting and calling need separate thinking
Many AI follow-up sequences mix calls and texts. Texts run on A2P 10DLC, registered through The Campaign Registry (TCR), with their own opt-in and STOP rules. Calls run on TCPA voice rules. A contact who replied STOP to your texts has not necessarily revoked call consent, and the reverse is also true, so track each channel on its own.
Can you use AI voice calls to reactivate old leads?
Only the leads whose consent covers an AI marketing call. Old lists often mix contacts who signed a written consent form with contacts who only called once or bought years ago. A reactivation call with an offer is telemarketing, so each number needs prior express written consent for that kind of call.
A safer pattern is to segment first. Contacts with documented written consent can receive AI calls. Contacts without it can get an email, or a text only where your SMS consent covers it, inviting them to opt in or book. That keeps the campaign useful without calling people who never agreed to an AI voice.
If your list is large and you are not sure what consent you hold, our GoHighLevel database reactivation service starts by sorting contacts by consent status before any outreach goes out.
Not sure your AI calling setup would hold up?
If outbound Voice AI is already live in your account, check it before a complaint does. The common gaps are aged lists without written consent, consent wording that never mentioned AI or prerecorded calls, opt-outs that do not reach Do Not Disturb, and agents that skip the business name.
These usually come from a quick build by a freelancer or an old snapshot imported without review. Our GoHighLevel sub-account audit reviews forms, consent fields, Do Not Disturb settings and Voice AI agents, and tells you what to fix. If you want the AI agents built properly from the start, our done-for-you GoHighLevel Voice AI and Conversation AI service configures inbound first and adds outbound only where your consent supports it. Plans are on our pricing page.
The TCPA allows damages of $500 per violation, and courts can raise that to $1,500 for willful or knowing violations. With per-call damages, a small list can add up quickly, so confirm your plan with counsel for your situation.
Frequently Asked Questions
Are AI voice calls legal in the US?
Yes, when they follow the TCPA. The FCC treats AI-generated voices as artificial voices, so outbound AI calls need the called person’s prior express consent, and marketing calls need prior express written consent. The calls must also identify the business, give a callback number, respect calling hours and honor Do Not Call requests.
Does the TCPA apply to AI voice calls?
Yes. In its February 2024 declaratory ruling, the FCC confirmed that the TCPA’s restrictions on artificial or prerecorded voice calls cover AI-generated voices, including voice cloning. An AI agent that can hold a conversation is still placing an artificial-voice call under the TCPA, so the same consent, identification and opt-out rules apply.
What consent do I need for AI sales calls?
Sales and promotional AI calls need prior express written consent. That is a signed agreement, which can be electronic, authorizing the business to make marketing calls using an artificial or prerecorded voice to that number. It cannot be required to buy anything. Keep a record of the form, the wording, the date and the source for each contact.
Do I need consent to use an AI receptionist for inbound calls?
The TCPA’s outbound consent rules do not work the same way for calls a customer places to you, because the customer started the call. Recording and transcription still raise consent questions in all-party consent states such as California and Florida. A short recording notice in the greeting is the common approach.
Does GoHighLevel check consent before Voice AI calls?
HighLevel’s current help center says the platform does not validate each contact’s consent before outbound Voice AI calls, and the business is responsible for consent. HighLevel adds guardrails such as a calling window, volume limits and an AI disclosure setting. Your forms, contact records and Do Not Disturb settings carry the compliance work.
What is the penalty for illegal AI robocalls?
The TCPA allows $500 in damages per violating call, and courts can increase that to $1,500 per call for willful or knowing violations. State attorneys general can also bring cases. Because damages count per call, an AI campaign to a list without proper consent can create large exposure even when the list is small.