Do You Have to Tell People They Are Talking to an AI? Disclosure Rules for AI Calls in the UAE, UK and US

Do you have to tell people they are talking to an AI?

Not under the three main texts we read. Rule. None of them contains a sentence that says an AI caller must announce that it is an AI. What each one does is regulate who may be called, with what consent, and what the caller must say about the business. Those duties still apply when the caller is an AI voice, and some of them are stricter for automated voices than for people.

What do the UAE rules say about an AI caller?

Rule. Cabinet Resolution 56 of 2024 regulates telemarketing, which it defines as phone calls a company makes to a consumer to market, advertise or promote a product or service, including marketing text messages. We read the English PDF from the Ministry of Economy and searched it for any clause about announcing that a caller is an AI or a machine. There is none. Article 5(6) says automated communication systems may be used for marketing under the resolution's provisions.

Rule. What Article 4 does require of marketing calls is relevant to an AI caller. Identify the company and the purpose of the call at the beginning (4(11)). Record the call and inform the consumer of the recording when the call begins (4(7)). Article 5(7) adds that the caller must ask the consumer whether they want to continue before starting to market. The fines for missing each of these are in our Cabinet Resolution 57 penalty table.

A call the customer makes to you, or a call you return at their request, is not the same thing. The resolution's definition of unwanted marketing calls leaves out marketing calls made at the request of the consumer. Our earlier posts on the 9-to-6 rule and on PDPL and call recording cover that split and the separate data-protection question. The UAE position we published before, that an agent confirms it is an AI if asked, is in can callers tell they are talking to an AI.

What do the UK rules say about an AI caller?

Rule. The Privacy and Electronic Communications Regulations 2003 do not mention AI. Two regulations matter. Regulation 19 bars recorded direct marketing messages sent through an automated calling system unless the subscriber has told the caller they consent to such calls on that line, and unless the caller does not hide its number or gives a line where it can be reached. It defines an automated calling system as one that can start a series of calls on its own and can play sounds that are not live speech. Regulation 21 stops unsolicited live marketing calls to anyone who has told the caller not to ring, or whose number is on the register kept under regulation 26. Regulation 24 says automated calls must give the caller's name and an address or a free phone number, and live calls must give the name and the address or number if asked.

Practice. We could not settle from the text alone whether a conversational AI that dials a list counts as an automated calling system under regulation 19 or as a live call under regulation 21. The ICO's telephone marketing page, which might say, blocked our reading with a 403. Until you have checked, treat an AI voice as automated: get the consent first, and give the name and a contact line. For a lead who filled in your form and agreed to be called, that consent should be on record. Our post on UK cold email rules covers the email side of the same regulations.

What do the US rules say about an AI caller?

Rule. The FCC's Declaratory Ruling FCC 24-17, adopted on 2 February 2024 and released on 8 February 2024, confirms that the TCPA's restriction on an "artificial or prerecorded voice" covers current AI technologies that generate human voices. Such calls need the prior express consent of the called party, unless there is an emergency purpose or an exemption. The ruling also says the rules on identification and disclosure for artificial or prerecorded voice messages apply, and that telemarketing messages must offer an opt-out method.

Rule. The disclosure that rule asks for is about the business. 47 CFR 64.1200(b) says the message must state the identity of the responsible business, individual or entity at the beginning, give its telephone number during or after the message, and for telemarketing offer an automated opt-out. Section 64.1200(a)(2) adds that a telemarketing call using an artificial or prerecorded voice needs the called party's prior express written consent. The eCFR text we read does not define AI.

Rule. State laws differ. California's bot law, Business and Professions Code section 17941, makes it unlawful to use a bot to communicate online with a person in California with the intent to mislead them about its artificial identity, to knowingly deceive them about the content of the communication in order to encourage a sale. Disclosing that it is a bot is a defence, and the disclosure must be clear and conspicuous. Section 17940 defines a bot as an automated online account and defines online by reference to public-facing websites and apps. We found nothing in that definition that reaches a phone call, but we did not research case law, and other states have their own AI laws that we did not read.

Does an AI receptionist have to say it is an AI?

Under the texts above, an AI receptionist that answers calls customers make to you is outside the marketing rules entirely, because those rules govern calls a business places to a consumer. Rule for the UAE: Resolution 56 defines telemarketing as calls a company makes. Practice: the rules that still matter for an inbound AI are data protection and recording. If you record the call, tell the caller, and keep your privacy notice current. The same applies to an AI that returns a call a person asked for.

What should an AI agent say at the start of a call?

Practice. Our advice is to say it, early and in plain words, even where no text we read requires it. There are three reasons. The business-identification and recording duties already force an opening statement, so adding one sentence costs little. A caller who asks "are you a person?" and gets a dodge learns that you will mislead them. And the law is moving: the FCC acted on AI voices in 2024, and California's bot rule turns on whether the user was misled. A short opening that does all of it looks like this:

"Good morning, this is an AI assistant calling for [business name]. Just so you know, this call will be recorded for quality and training purposes. I'm calling about the enquiry you sent today. Is now a good time to continue?"

Illustrative opening, not a recording of a real call

That line covers Resolution 56's company-and-purpose, recording notice and continue-or-not questions in the UAE, the name-and-purpose opening in the UK and US, and the plain statement about AI. You still need to give the phone number (US) or an address or free number (UK) during or after the call. If the person asks to speak to a human, the agent should hand over or arrange a call back. We would also never have an agent deny being an AI when asked.

What does this mean for MAJ Leads?

AI calling is coming soon in MAJ Leads, and every MAJ agent is built to open with "Just so you know, this call will be recorded for quality and training purposes." Today MAJ Leads is a lead management system: your own reps make the calls, and a call is due for the assigned rep within about 30 seconds of a lead arriving. See how it works and pricing. We have no results from AI calls to report.