No neutral study measures whether renters can tell they're talking to an AI voice agent — that number isn't public. So judge it on a test call: latency, whether you can interrupt it, and what it does with an off-script question. And in Utah, it has to tell you when you ask.
Is There a Published Number for How Often Renters Can Tell?
No. Nothing publicly measures detection rates on a live call, and it would be easy to overstate one by citing familiarity instead. What Pew Research Center does track is usage: 49% of U.S. adults said they had ever used an AI chatbot in a survey conducted February 17–23, 2026 (Pew Research Center, "Americans and AI 2026"). That measures how common AI interaction has become, not whether a caller can pick an AI voice out mid-conversation — different questions, and only the first has a public number.
What Should You Listen For on a Test Call?
Call the line yourself and run it at the seams, not down the middle. Six things separate a system that converses from one that reads a script:
- Latency, and whether it changes. An even gap before every reply reads as human. A gap that stretches on the harder question gives it away.
- Interruption. Talk over it mid-sentence. Does it stop and follow you, or finish the sentence it had queued?
- An off-script question. Ask something nobody briefed it on — a neighbourhood question, a lease oddity. Improvising is worse than admitting the gap.
- The handoff. Ask for a person. A clean system says what it's doing, and what you told it survives the transfer.
- Rough input. An accent, a spelled-out street name, a dog barking. Repetition and mishearing surface here first.
- Repetition. Ask the same thing twice, worded differently. Word-for-word identical answers are the tell.
LetHub's AI voice agent answers rental inquiries by phone, text, and chat 24/7, and there's a live line anyone can dial to run that test: 404-383-6213. Better than any vendor's claim, this one included — or see how it handles a real call.
Does the Law Require an AI Phone Agent to Announce Itself?
In at least one state, yes — when the caller asks. Utah Code § 13-75-103, enacted by S.B. 226 and effective May 7, 2025, requires a supplier that uses generative AI to interact with someone in a consumer transaction to disclose that the person "is interacting with generative artificial intelligence and not a human, if the individual asks or otherwise prompts the supplier about whether artificial intelligence is being used" — the question must be clear and unambiguous. Utah's definition covers a system designed to simulate human conversation through text, audio, or visual communication, so a spoken call is inside it. Regulated occupations carry a stronger, unprompted duty for high-risk interactions — disclosure "verbally at the start of a verbal interaction" — and § 13-75-104 is a safe harbour for disclosing clearly up front (Utah S.B. 226, enrolled).
Two other statutes people reach for don't cover an inbound call. California's bot statute makes it unlawful to use a bot to communicate "with another person in California online" with intent to mislead about its artificial identity in a commercial transaction (Cal. Bus. & Prof. Code § 17941) — scoped to online communication by its own text. And the FCC's February 2024 ruling confirms AI-generated voices count as "artificial" under the Telephone Consumer Protection Act, but that provision restricts calls a business initiates with a synthetic voice, not a call a renter places to you (FCC 24-17).
Three authorities, three different answers. There is no single national disclosure rule, and one state already requires an answer on request. The practical default: answer honestly the moment a caller asks. This is general information, not legal advice — confirm the states you operate in with counsel.
Frequently Asked Questions
Can renters tell when they're talking to an AI voice agent?
No published, neutral measurement answers that directly. What Pew Research tracks is usage — 49% of U.S. adults said they had ever used an AI chatbot in February 2026 — which shows AI interaction is common, not that a caller can identify one mid-call. Judge a specific system by calling it.
How do I test whether an AI phone agent sounds human?
Call it and push on six things: how long it waits to reply and whether that gap changes, whether you can interrupt it mid-sentence, what it does with an off-script question, how a transfer sounds, how it handles noise or an accent, and whether repeated questions get identical answers.
Does any law require an AI phone agent to say it's an AI?
Utah does, on request: § 13-75-103 requires a supplier using generative AI in a consumer transaction to disclose it when the individual clearly asks, and Utah's definition expressly covers audio conversation. California's bot statute is limited to online communication, and the FCC's AI-voice ruling addresses outbound calls, not inbound answering.
Does the FCC require disclosure that a call is AI-generated?
The FCC's February 2024 ruling addresses outbound calls a business initiates using an AI-generated voice under the TCPA. It doesn't impose a duty on an AI agent to identify itself when answering an inbound call a renter placed to your business line. State law is where an inbound duty can arise.
