Federal Communications Commission

The Federal Communications Commission (FCC) is the US agency that administers the Communications Act of 1934, as amended.^[1] That statute is the framework the commission works from: it sets what the agency may regulate and what it may not, and when Congress amends it, the commission's rules shift with it. Much of the day-to-day business is unglamorous — licensing, spectrum, the rules that govern networks and callers — but the agency also decides what counts as an abusive or illegal call, which is how it came to have a say over AI voices.

The ruling on AI voices

On 8 February 2024 the FCC adopted a declaratory ruling confirming that restrictions on calls using an "artificial or prerecorded voice" already encompass AI technologies that generate artificial human voices.^[1] A declaratory ruling does not create a new obligation; it states how the commission reads a rule that already exists. That is why the agency could address voice cloning without waiting for wording aimed specifically at AI.

The rule being interpreted comes from consumer protection law: telemarketers generally cannot use automated dialers or artificial or prerecorded voice messages to call cellphones.^[2] The concern the commission named was not the technology in itself but what a cloned voice can accomplish. The FCC said voice cloning can convince a called party that a trusted person, or someone they care about, wants or needs them to take some action they would not otherwise take.^[1] The deceptive force comes from familiarity — a voice the listener recognises is doing the asking.

The New Hampshire robocalls

The ruling arrived alongside an enforcement case. The Commission's Enforcement Bureau investigated the robocall campaign that targeted New Hampshire voters before the 2024 primary.^[1] The commission proposed a US$6 million forfeiture — a monetary penalty — against Steve Kramer over that campaign.^[1] It found that Kramer knowingly caused thousands of illegal prerecorded voice robocalls using misleading and inaccurate caller ID information.^[1] The two strands of the announcement fit together: a statement of what the rules already cover, and a case applying them.

Penalties and legal routes

The commission has spelled out the exposure for violators: those who break the law can face fines of more than US$23,000 per call.^[2] The ruling also widened the routes to a remedy. It lets call recipients file lawsuits, and it gives state attorneys general a new mechanism to crack down on violators.^[3] Robocall enforcement did not start with AI. In 2023 the FCC finalised a US$5.1 million fine against conservative activists for making more than 1,100 illegal robocalls ahead of the 2020 US election.^[3]

Where it fits

For an agency that administers a statute from 1934, the AI voice question was a test of how far old language stretches. The commission's answer — that a synthetic human voice is an "artificial" voice — placed cloned speech inside an existing consumer protection regime rather than outside it. The practical consequence is that the same complaint channels, the same per-call penalties and the same private lawsuits now reach calls made with generated voices.

The ruling does not settle every question about synthetic media. It addresses calls, and in particular calls made with an artificial or prerecorded voice; other uses of voice cloning fall outside what it decided. What it does establish is a position other regulators, courts and callers can argue from, and it makes the commission one of the first agencies to say plainly that a machine-generated human voice is covered by rules written long before such voices existed.