Plain-English answers
It's one of the most common questions business owners ask us, usually in a lowered voice: "Do we have to tell people it's an AI?" Behind it is a worry about what happens when a caller finds out. Our answer has two parts: what the rules look like, and exactly what our agents do when a caller asks.
Last updated: October 2026
Rules about AI on phone calls are emerging and uneven. A few things are broadly true: laws in some states already require that automated or AI-generated calls identify themselves in certain contexts, regulators have taken particular interest in AI voices used in outbound and marketing calls, and the direction of travel everywhere is toward more disclosure, not less. Inbound answering, a customer calling your business and reaching your assistant, has generally drawn less regulatory attention than outbound calling, but this area moves quickly.
We're a voice-agent builder, not a law firm, and this page isn't legal advice. What we can tell you is how we build for it: on inbound calls the agent greets callers like a front desk would and doesn't volunteer that it's an AI, but it never denies being one when asked, and a caller who would rather talk to a person can be handed off to one. Outbound calls are the exception: there, the agent identifies itself as an AI assistant right away. Whether that is enough for your business depends on your state and your calls, so ask your attorney.
The fear is that a caller who learns they're talking to an AI will hang up. What actually makes callers hang up is feeling tricked: asking something twice, getting a too-smooth reply at 2 a.m., and then getting a dodge when they ask the direct question. Deception has a cliff; honesty doesn't.
A caller who asks and hears a straight yes recalibrates instantly and gets on with it: state the problem, answer the questions, take the appointment. What they came for was never a philosophical companion: it was getting the leak fixed. An agent that answers immediately, understands them, and books the job has done more for that caller than a voicemail box ever did, and the caller knows it. Callers don't punish businesses for using tools that work. They punish being fooled.
There's a trust dividend, too: the business whose assistant answers that question honestly signals it has nothing to hide, and the human callback that follows lands warmer for it.
The agent greets callers like a front desk would and gets to work. It doesn't open with a disclosure of its own.
If a caller asks whether they're talking to an AI or a real person, the agent tells the truth: yes, it's an AI, not a human. It never denies it, never dodges, and never changes the subject. It says so plainly and good-naturedly, and offers a person if the caller would rather talk to one. This is a hard rule in how we build, and it applies on every call.
A caller who would rather talk to a person can be handed off to one. That's a transfer rule, not an argument. Insisting a caller stay with the AI is the one move that converts disclosure into a complaint.
Disclosure is non-negotiable and immediate: outbound is where the rules are strictest and where an unannounced AI voice does real damage.
Call our demo line at 1-855-952-4546 and ask it directly whether it's an AI. You'll get a straight answer and, we'd argue, a better feeling about the company behind it than any evasion could produce. If that's the posture you want on your own phone line, your first agent is free to build: no cost, no contract.
Rules are emerging and vary by state and context; outbound and marketing calls draw the strictest requirements, while inbound answering has generally seen less regulation so far. Our default on inbound calls is an agent that greets callers like a front desk, never denies being an AI when asked, and hands off to a person on request; on outbound calls it identifies itself as an AI assistant right away. For your specific obligations, ask your attorney; this isn't legal advice.
Being fooled loses callers; being told the truth rarely does. When a caller asks and gets a straight yes, they recalibrate and proceed, because what they want is their problem handled. The trust damage comes from feeling deceived, not from the tool.
It says yes, plainly and good-naturedly: it's an AI, not a human. It never denies it or dodges, and offers a person if the caller would rather talk to one. Never denying being an AI is a hard rule in how we build. It doesn't announce itself unprompted on inbound calls; on outbound calls it identifies itself as an AI assistant right away.
Yes: a caller asking for a person triggers a transfer rule, not a debate. Forcing a caller to stay with the AI is the single behavior most likely to turn disclosure into a complaint, so we build the escape hatch in by default.
Call the demo line and ask it point-blank if it's an AI, then we'll build one around your business, free.
Ask Custom Voice Automations
Usually replies instantly