There's no single federal law in the U.S. requiring a business to tell callers they're speaking with an AI system, but the legal picture varies by state and is changing quickly. Some states have passed or proposed laws requiring disclosure in specific situations, like automated telemarketing calls or interactions that could reasonably be mistaken for a human when it matters, such as medical, legal, or financial advice.
Because rules differ by state and continue to evolve, a business operating in multiple states should check the specific requirements where its callers are located rather than assume one national standard applies. A handful of states, California among them, have moved fastest on AI disclosure and identity rules in various contexts, and more are expected to follow. Even where it isn't legally required, most businesses configure their AI phone system to identify itself honestly if a caller asks directly, since being upfront doesn't meaningfully hurt the call. Callers overwhelmingly care about getting a fast, accurate answer and a resolved problem; whether the voice on the other end is human rarely changes that outcome once the call goes well. In practice, this usually means a short line built into the AI's script, something like confirming it's an automated assistant when a caller asks, and offering to connect to a person if the caller would prefer one. That single sentence covers the disclosure question in nearly every jurisdiction and avoids the goodwill hit of a caller feeling misled after the fact.
Key takeaways
- No federal law in the U.S. requires disclosing that a caller is speaking with AI, but state laws vary and are changing quickly.
- States that regulate AI disclosure usually target specific situations, like automated telemarketing or contexts where a caller could be misled about who they're dealing with.
- A business should check the rules in the states where its actual customers are located rather than assume one nationwide standard applies.
- Configuring the system to honestly identify itself as AI when a caller asks directly is considered best practice even where it isn't legally required.
Answered by Alex Rivera, Founder · Updated July 24, 2026