AI voice calls
AI robocall / voice cloning / synthetic voice / AI outbound calling
An AI voice call uses a synthesized or cloned voice to speak to the person who answers. In February 2024 the FCC confirmed those voices fall under the existing rules on artificial or prerecorded voice, so the same consent requirements apply. The message must also identify your business and offer an automated way to opt out.
The most useful thing about the ruling is what it did not do. It wrote no new rulebook for AI. In February 2024 the Commission confirmed that the existing restrictions on artificial or prerecorded voice cover current AI technologies that resemble human voices. The consequence follows automatically. Callers must obtain prior express consent before making such a call. If the call carries an advertisement or amounts to telemarketing, the rules require prior express written consent.
So the question to ask about an AI calling tool is not whether regulation exists yet. It is whether you would have been allowed to make the same call with a recording, because the regulator has put the two in one category.
The obligations that follow are specific, and they are script requirements rather than settings. Under 47 CFR 64.1200, every artificial or prerecorded voice message must open by stating clearly who is responsible for the call. For a business, that means the name it is registered under with the state corporation commission or a comparable authority.
The message must also state a telephone number for that business, during or after it. That number may not be a 900 number, or any number charged above normal rates. For telemarketing to a residential line, it has to let a person make a do-not-call request during business hours. Such calls must also carry an automated opt-out, by voice or key press, inside the call.
In practice
Those requirements sit awkwardly with the thing AI voice tools are sold on, which is sounding natural and improvising. A conversational agent still has to say the registered business name at the beginning, every time, and still has to offer a working automated opt-out during the call. Those are not tone choices. A demo that sounds impressive and skips them is demonstrating a call you could not lawfully place.
Not the same as
- An AI receptionist answering your line
- These rules attach to the party responsible for initiating a call. Answering one your customer placed is a different situation.
- Call recording
- Whether a call may be recorded is a separate question with its own consent rules.
Why it matters to you
AI calling is sold as a new category, and the regulator has said plainly that it is an old one. That is good news for anyone evaluating a tool. The rules are already written, already specific, and testable against a script. The wrong question is whether AI voice is allowed. The right one is simpler. Would this call, to this number, with this script, have been allowed as a recording?
What to ask or check
- 01Do we hold prior express written consent for every number this will dial, if the call advertises anything?
- 02Does the script state our registered business name at the very beginning of the message?
- 03Is there an automated opt-out inside the call, and has anyone tested that it works?
What people get wrong
That AI voice calling is ahead of the rules. The FCC confirmed in February 2024 that the existing artificial or prerecorded voice restrictions already cover AI voices that resemble human speech.
Red flags
- A vendor describing AI voice as a regulatory grey area.
- A script that does not open with the registered business name.
- No automated opt-out inside the call, or one nobody has tested.
Who owns it
The party responsible for initiating the call. Buying the technology from somebody else does not move that.
Where you will see it
In pitches for AI outbound calling and voice agents, and in any proposal to follow up leads automatically by phone.