Imagine a sign that has to be hung on the door before the conversation starts: “you're talking to a robot.” Some states require that sign in certain situations (mostly selling or political messaging), some don't require it yet, and the list grows almost every month. Because the rules differ, the safe operating default is to disclose the AI clearly at the start of every call unless you have a specific reason not to.
The notable state laws today
| Jurisdiction | Law | Gist |
|---|---|---|
| California | B.O.T. Act (SB 1001) | Must disclose use of a bot when used to incentivize a sale or influence a vote in certain commercial/political contexts. |
| Utah | AI Policy Act (S.B. 149) | Must disclose generative-AI use in regulated occupations, and on a consumer's request in others. |
| Colorado | Colorado AI Act (SB 24-205) | Duties for high-risk AI systems and consumer-facing transparency (effective 2026). |
| EU (not a state) | EU AI Act, Art. 50 | People must be told when they interact with an AI system unless it's obvious. |
New state AI bills appear constantly, and many target voice/calling specifically. Treat any static list — including this one — as a snapshot. The interactive state tracker on the hub is the version we keep current.
What a good AI disclosure sounds like
It should be early, plain, and unmissable — not buried in legalese. A workable pattern: “Hi, this is an AI assistant calling on behalf of [company]. This call may be recorded. How can I help?” That single line can satisfy an AI-disclosure requirement and a recording-disclosure requirement at once.
Writing the disclosure into the script is step one; proving the agent actually said it on every call is step two. Trace checks each call for the AI-disclosure line by jurisdiction and flags the turns that miss it — so a script change that quietly broke the disclosure doesn't become a thousand violations.
Frequently asked questions
Is there a federal law requiring AI call disclosure?+
No single federal statute requires announcing an AI voice today. Disclosure obligations come mainly from state laws (like California's B.O.T. Act) and, internationally, the EU AI Act. The federal TCPA governs consent and opt-outs rather than AI disclosure per se.
Which states require bot/AI disclosure?+
California (B.O.T. Act), Utah (AI Policy Act), and Colorado (Colorado AI Act) are leading examples, with more states adding requirements regularly. Scope and triggers differ by state and use case, so verify the current rule for each state you call into.
What should my AI agent say to disclose itself?+
A clear, early statement that the caller is an AI acting for a named company, for example, 'This is an AI assistant calling on behalf of [company], and this call may be recorded.' Keep it plain and unmissable.
Primary sources
We cite primary law. Statutes, rulings, and state laws change, confirm currency before relying on them.
- California B.O.T. Act, SB 1001 (Cal. Bus. & Prof. Code § 17940 et seq.)
Requires disclosure when a bot is used to incentivize a sale or influence a vote in some contexts.
- Utah AI Policy Act (S.B. 149)
Disclosure obligations when generative AI is used in regulated occupations / on request.
- Colorado AI Act (SB 24-205)
Obligations for developers/deployers of high-risk AI systems (effective 2026).
- EU AI Act (Regulation (EU) 2024/1689)
Transparency duties for AI systems that interact with people (Art. 50).