Are Voice Bots Legal? Disclosure Rules and Regulations

The FCC's 2024 Ruling on AI Voices and the TCPA

On February 8, 2024, the FCC adopted a Declaratory Ruling, FCC 24-17, clarifying that calls using AI-generated or voice-cloned technology count as "artificial voice" calls under the Telephone Consumer Protection Act. The ruling was prompted by a January 2024 incident in which a robocall used an AI-cloned voice of a public figure to discourage people from voting. It's worth being precise about what the FCC actually did here: it did not write a brand-new AI-specific law. It clarified that the TCPA's existing rules for prerecorded and artificial-voice calls already cover calls generated or cloned with AI. Practically, that means a voice AI agent calling someone is treated the same way an old-style prerecorded robocall would be treated under the TCPA, and it needs to meet the same consent requirements. The FCC's own announcement of this ruling is titled "FCC Makes AI-Generated Voices in Robocalls Illegal," but the more accurate way to read that headline is "AI-generated voices in robocalls are illegal when they don't have the consent the TCPA already requires," not that AI voice calls are illegal outright.

California's Bot Law (and Why It Doesn't Cover Phone Calls)

California's B.O.T. Act, SB 1001, effective since July 1, 2019, often comes up in searches about "bot disclosure laws," so it's worth addressing directly. The law makes it unlawful to use a bot to communicate online with a person with the intent to mislead them about the bot's artificial identity, when that communication is meant to incentivize a sale or influence a vote, unless the bot maker gives clear and conspicuous disclosure. Two scope limits matter a lot here. First, it only applies to large online platforms, defined as those with 10 million or more unique monthly United States visitors, not to smaller businesses running their own bots. Second, and more importantly for voice AI, the law is about online bots, meaning chatbots and automated accounts on websites, apps, and social platforms. It does not, on its own text, regulate phone calls. If your voice agent is calling people on the phone rather than chatting with them through a website or app, SB 1001 is not the law you need to be checking. It's frequently cited in articles about "AI disclosure law" in a way that blurs this distinction, so we want to be clear about it here.

Utah's AI Policy Act: The Law That's Actually About Voice Disclosure

Utah's Artificial Intelligence Policy Act, effective May 1, 2024, is a more directly relevant law for anyone running voice AI agents, according to a legal analysis published by the law firm Davis Wright Tremaine. As originally enacted, it requires anyone providing services in a "regulated occupation" (fields like healthcare, law, and other licensed professions) using generative AI to prominently disclose, verbally, at the start of the conversation, that the person is interacting with AI rather than a human. For businesses outside those regulated occupations, the original law only required disclosure if the person asked whether they were talking to a bot. The law was later amended by SB 226 to tighten this: for "high-risk" interactions, meaning conversations touching on health, financial, biometric, or legal advice, disclosure is now required upfront regardless of whether the customer asks. Utah is not likely to be the last state to legislate specifically on voice and chat AI disclosure; more states have introduced similar bills, and the honest expectation for anyone building or deploying voice agents is that this list of state-specific disclosure rules will keep growing over the next few years.

How This Ties Back to the Existing TCPA Consent Framework

Underneath all of this sits a consent framework that predates the 2024 AI ruling and applies independently of it. Under the TCPA, prerecorded or artificial-voice calls to residential landlines generally require prior express consent from the person being called. Calls to wireless numbers for marketing purposes require a stricter standard: prior express written consent. The FCC's 2024 ruling didn't change these underlying consent thresholds; it confirmed that AI-generated voices fall inside this framework rather than existing outside it in some regulatory gray zone. That's the most important practical takeaway from the ruling: if your organization already has a compliant process for prerecorded or robocall consent, that same process is the right starting point for a voice AI agent making outbound calls. If you don't have that process yet, building one, and documenting how and when consent was obtained, is foundational compliance work regardless of whether you're using a script, a recording, or an AI agent.

Practical Guidance for Teams Deploying Voice Agents

Given all of the above, the practical path for most teams is straightforward even though the legal landscape is fragmented. Disclose that a caller is talking to an AI system when a law in your jurisdiction or industry requires it, and consider doing so even when it isn't strictly required, since it tends to build trust with callers rather than undermine it. Get and document proper consent before placing outbound calls, following the TCPA's existing standards as a baseline regardless of whether AI is involved. Pay close attention to your specific industry: healthcare, financial services, and legal services are the areas where regulators, including Utah's amended law, have singled out AI disclosure as non-negotiable. Track state-level developments, since California and Utah are unlikely to be the only states with rules in this space by the time you read this. At Persistence, we build our voice agent platform with these kinds of consent and disclosure requirements in mind, but the specific rules that apply to your calls depend on your jurisdiction, your industry, and how you're using the technology. None of this article is legal advice, and it shouldn't be treated as a compliance checklist for your business. Talk to a lawyer who can look at your specific use case, the states you're calling into, and the laws that actually apply before you launch.

Key takeaways

• Yes, voice bots are legal to operate, but disclosure and consent requirements apply, and they vary by jurisdiction, industry, and how the call is used.
• In February 2024, the FCC ruled that AI-generated and voice-cloned calls count as "artificial voice" calls under the TCPA, so the same prior-express-consent rules that already applied to robocalls now clearly apply to AI voice calls too.
• California's bot disclosure law (SB 1001) applies to large online platforms communicating with users online, not to phone calls, so it does not directly govern voice AI agents making calls.
• Utah's AI Policy Act is more directly on point: it requires certain businesses, especially those in regulated occupations or high-risk interactions, to verbally disclose that a caller is talking to AI.
• This article is a general educational overview, not legal advice. Confirm requirements for your specific jurisdiction, industry, and use case with a qualified attorney before deploying a voice agent.