# An AI answered the phone: how do I reach a person, and should I let an AI call for me?

Phrases and steps that get you past an AI receptionist to a person, what to get confirmed in writing, where the law says a bot must say it is a bot, and a short approval list if you let your phone's AI make a call for you.

Guides · v1.0 · draft · https://hayzlabs.com/guides/when-an-ai-answers-the-phone/

In short:
- Ask plainly for a person, ask whether you are talking to an AI, and switch to chat or email if the line keeps looping.
- Before you hang up on an AI receptionist, get a reference number and a written confirmation of anything agreed.
- If an AI calls for you, approve only what the call needs. Never let it read out card numbers or one-time codes.

> [!NOT ADVICE]
> This page is general information, not legal or financial advice. Anything you let an AI say on a call can be heard and recorded by the other side, so treat payment details and security codes as off limits.

## Short answer

When an AI answers a business line, ask in plain words for a person, and ask whether you're speaking to an AI. In the EU, the AI Act says people must be told when they're interacting with an AI system unless it's obvious, and that rule applies from 2 August 2026, according to the [European Commission](https://digital-strategy.ec.europa.eu/en/faqs/transparency-obligations-under-article-50-ai-act) (regulator guidance). If the line loops, switch to chat, email or a different number, and get anything agreed confirmed in writing. Letting your phone's AI call a business for you is still an early experiment as of October 2026: approve only what that call needs, and read the transcript afterwards.

If the call is about money, a family emergency or a code, it may be a scam rather than a receptionist. Use [AI voice scams: hang up and call back](https://hayzlabs.com/guides/hang-up-and-call-back/) instead.

## Why asking for a person is reasonable

You're not alone in wanting one. In an analyst survey of 3,566 business and consumer customers in February and March 2026, "87% of customers say access to a human agent is essential" when companies use generative AI in customer service, and only 27% said they would try a chatbot again after a bad experience ([Gartner, 2 September 2026](https://gcom.pdo.aws.gartner.com/en/newsroom/press-releases/2026-09-02-gartner-finds-only-27-percent-of-customers-would-try-a-chatbot-again-after-a-negative-experience), analyst survey about chatbots in general, not phone systems). Gartner's own advice to companies is that chatbots should be "seamless connectors to human support, not containment traps."

## Get a person: the card

> [!OUR RULE]
> These phrases and the order are our suggestion. No study we've read tests which wording works best, and every company's system is different.

```text title="Get a person · v1.0"
1. "Am I speaking to an AI or a person?"
2. "I'd like to speak to a person, please."  (Say it again, calmly, if needed.)
3. If it asks why: give one short reason, such as
   "It's a billing dispute" or "It's about a complaint."
4. On a keypad menu, try the option for an existing account,
   a complaint or "something else". Some lines connect you if you press 0.
5. If it loops three times: hang up and try the chat on the website,
   email, or the number on your bill or contract.
6. Before you hang up, every time:
   "What is the reference number for this call?"
   "Please send me that in writing, by email or text."
```

We share the "ask for a person" idea with [AI job interviews](https://hayzlabs.com/guides/interviewed-by-ai/), where the same plain request applies to an AI interviewer.

## Get it confirmed in writing

An AI receptionist can book, cancel or promise things just as a person can, and it can also get them wrong. Before the call ends, ask it to confirm:

- **What was agreed**: the booking, change, refund or cancellation, in one sentence.
- **The date and time** it takes effect.
- **A reference number** you can quote later.
- **Where the confirmation goes**: your email or a text, sent now.

If nothing arrives within the day, call back or write in with the reference number. Keep the email or a screenshot. That's your proof if the business later says nothing was agreed (our suggestion).

## Where the law says a bot must say it's a bot

- **EU.** Article 50(1) of the AI Act says AI systems "intended to interact directly with natural persons" must be built so that people "are informed that they are interacting with an AI system", unless that is obvious to a reasonably well-informed person in the context ([AI Act Service Desk, Article 50](https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50), law text). The Commission says the rule "applies as from 2 August 2026", lists chatbots and AI agents as examples, and says people must be told "from the start of the first interaction" ([European Commission FAQ, updated 24 July 2026](https://digital-strategy.ec.europa.eu/en/faqs/transparency-obligations-under-article-50-ai-act), regulator guidance). The FAQ doesn't mention phone calls by name.
- **Utah, US.** A 2025 Utah law says a business that uses generative AI (AI that writes or speaks new content) to deal with you in a consumer transaction must tell you that you're "interacting with generative artificial intelligence and not a human" if you ask. Your question has to be "a clear and unambiguous request" ([Utah SB 226, enrolled 2025](https://le.utah.gov/Session/2025/bills/enrolled/SB0226.pdf), law text). That's one reason the card starts with a direct question.
- **Elsewhere.** We have only checked these two. Other US states and countries have their own rules, and some apply only online. Check your own state or country's consumer protection office.

## Letting an AI call for you

As of October 2026, Google is testing "Call for Me", in which Gemini phones a business for you. It's an experiment for Pixel 11 owners in the US who pay for a Gemini subscription, through the beta Phone app. According to [TechCrunch (24 September 2026)](https://techcrunch.com/2026/09/24/google-tests-letting-gemini-make-phone-calls-initially-for-us-pixel-owners/) (news), it calls from your own number, finds its way through phone menus, waits on hold, can share personal details you approve, and shows a live transcript you can "take over at any time". The report doesn't say whether it tells the business it's an AI, so don't assume either way.

If you try it, or a tool like it, this is the approval list we suggest:

```text title="Before an AI calls for me · v1.0"
This call is to: <ONE_TASK>
It may share:     my name, <CALLBACK_NUMBER>, <BOOKING_DETAILS>
It may not share: card numbers, bank details, one-time codes, passwords,
                  my address or date of birth, unless I take over the call.
It may not agree to: payments, deposits, cancellation fees or new contracts.
After the call I will: read the transcript and check for a written confirmation.
```

Replace the following:
- `<ONE_TASK>`: a single errand, such as `book a table for four on Friday at 7pm`.
- `<CALLBACK_NUMBER>`: the number the business should use to reach you.
- `<BOOKING_DETAILS>`: only what the errand needs, such as party size or an existing booking reference.

If the business asks for a card, a code or anything off the list, take over the call or end it and phone yourself. Don't paste account numbers or codes into the AI's chat to "help it along" ([What not to paste](https://hayzlabs.com/start/what-not-to-paste/)).

> [!CHECK]
> It worked if you reached a person or got your answer, you have a reference number and a written confirmation, and, if an AI called for you, the transcript shows it shared nothing outside your approval list.

## Limits

> [!LIMIT]
> Phone systems differ by company, so no phrase is sure to work. The legal section covers only the EU AI Act and one Utah law, read on official pages on 19 October 2026; it isn't a full list and laws change. "Call for Me" is an early, US-only experiment described by one news report as of October 2026, and we haven't used it. The Gartner figures are about chatbots in general and come from an analyst firm that advises companies.

## Sources

1. European Commission. Transparency obligations under Article 50 of the AI Act (FAQ), updated 24 July 2026. Regulator guidance. [Link](https://digital-strategy.ec.europa.eu/en/faqs/transparency-obligations-under-article-50-ai-act)
2. European Commission AI Act Service Desk. Article 50: Transparency obligations for providers and deployers of certain AI systems, consolidated text as at 27 July 2026. Law text. [Link](https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50)
3. Utah Legislature. SB 226, Artificial Intelligence Consumer Protection Amendments, enrolled 2025. Law text. [Link](https://le.utah.gov/Session/2025/bills/enrolled/SB0226.pdf)
4. Gartner. Gartner finds only 27 percent of customers would try a chatbot again after a negative experience, 2 September 2026. Analyst survey. [Link](https://gcom.pdo.aws.gartner.com/en/newsroom/press-releases/2026-09-02-gartner-finds-only-27-percent-of-customers-would-try-a-chatbot-again-after-a-negative-experience)
5. TechCrunch. Google tests letting Gemini make phone calls, initially for US Pixel owners, 24 September 2026. News. [Link](https://techcrunch.com/2026/09/24/google-tests-letting-gemini-make-phone-calls-initially-for-us-pixel-owners/)