Is AI Voice Calling Legal? US, UK & UAE Rules Explained

TL;DR: Using an AI voice agent is legal in the US, UK, and UAE. Using one to cold-call strangers who never asked to hear from you is restricted almost everywhere — and the restriction has little to do with the "AI" part. So is AI voice calling legal? Yes, when the law's three questions go your way: who placed the call, is there consent, and do you say who's calling. Answer your own inbound calls and return calls to people who just contacted you, and you're on the safe side of every rule below.

Most of the panic about whether AI calling is "allowed" conflates two completely different things: an AI agent that answers your phone and an AI that dials ten thousand strangers. Regulators draw a hard line between them. The line isn't "human vs robot" — it's "did the customer reach out, or did you interrupt their dinner." Get that distinction right and the compliance picture gets a lot simpler. This is a plain-English overview, not legal advice; confirm the specifics for your market with counsel before you launch outbound campaigns.

Is AI voice calling legal?

Yes — operating an AI voice agent is legal in all three of Heysav's markets. What's regulated is outbound marketing to people who haven't consented, and that has been regulated for decades regardless of whether a person or a machine makes the call. The "artificial voice" simply slots into existing telemarketing law. Here's how each market treats it.

Jurisdiction	Governing rules	Outbound cold AI calls to strangers	Answering inbound / requested callbacks
US	TCPA + FCC Declaratory Ruling (FCC 24-17)	Restricted — prior express consent required	Not a "robocall"; allowed
UK	PECR (+ UK GDPR)	Restricted — automated calls need consent	Solicited/requested calls largely exempt
UAE	Cabinet Resolutions 56 & 57 of 2024 (TDRA)	Heavily restricted — approval + DNCR + hours	Inbound isn't "telemarketing"

Is AI voice calling legal in the US under the TCPA?

Legal, but cold outbound calls with an AI voice need consent. On February 8, 2024, the FCC adopted a Declaratory Ruling recognizing that calls made with AI-generated voices are "artificial" under the Telephone Consumer Protection Act (TCPA). In plain terms: an AI voice is treated exactly like a prerecorded robocall.

What the ruling actually requires comes straight from the text. The TCPA prohibits initiating "any telephone call to any residential telephone line using an artificial or prerecorded voice to deliver a message without the prior express consent of the called party" unless an exception applies, and the FCC confirmed these restrictions encompass current AI technologies that generate human voices — applicable to any AI technology that initiates any outbound telephone call. If the call is an advertisement or telemarketing, the bar is higher — the caller must obtain prior express written consent, provide identification, and offer an opt-out.

Read the verbs carefully: the law restricts initiating and making outbound calls. It is not a ban on the technology — see our reality check on whether AI voice agents actually sound human — it's a ban on dialing people who didn't agree to be dialed. That single word, "initiate," is why answering inbound calls sits in a different category entirely.

Is AI voice calling legal in the UK under PECR?

Yes, with a sharp split between live and automated calls. Under the Privacy and Electronic Communications Regulations (PECR), for most types of live call you can make unsolicited marketing calls if the person has not objected and the number is not on the Telephone Preference Service (TPS) or, for businesses, the Corporate TPS (CTPS) — and a registration takes 28 days to take effect. You must also display your number, say who you are, and give contact details on request.

Automated calls are the strict ones, and an AI voice agent arguably looks more like an automated system than a live human. As the UK's data regulator puts it, an organization is only permitted to make an automated call — one made by an automated dialing system that plays a recorded message — if the person has specifically consented to receiving automated calls from that organization. That consent rule applies whether you're calling an individual or a business. The safe assumption for any outbound AI marketing call in the UK is that the stricter, consent-first standard applies.

The relief valve is "solicited." PECR restricts unsolicited marketing — if someone specifically asks you to send them information, you can do so without worrying about most of the rules. A customer who asks you to call them back has solicited that call. Enforcement is real: the ICO can issue fines of up to £500,000 for serious breaches, and one company that made over six million recorded marketing calls without prior consent was penalised £200,000.

Is AI voice calling legal in the UAE under the TDRA rules?

Yes for legitimate, approved use — but the UAE has the most prescriptive cold-calling regime of the three. Cabinet Resolution No. 56 of 2024 concerning the telemarketing regulations came into effect on 27 August 2024, paired with Resolution 57 on penalties. The obligations are concrete:

• Prior approval and training. Businesses must obtain prior approval from the Telecommunications and Digital Government Regulatory Authority (TDRA) to conduct telemarketing, and train staff on the Do Not Call Register.
• Local numbers, calling hours, and consent. UAE telemarketers must use local numbers registered under the company's licence, keep records of marketing calls, only call between 9am and 6pm, confirm the customer wants to continue before promoting anything, and avoid unjustified pressure.
• Transparency. Telemarketers must introduce the company name and the purpose of the call at the beginning, and the rules also cover marketing SMS and social-media messages.
• The Do Not Call Registry. Consumers can register on the DNCR to block marketing calls, and contacting them anyway is a violation.

The teeth: companies that fail to comply could face fines of up to Dh150,000 (about US$40,840), plus warnings, suspension of telemarketing activity, and potentially revocation of the licence and cutting of communication services. Separately, data handling falls under the UAE Personal Data Protection Law (Federal Decree-Law No. 45 of 2021), so consent and the source of your customer data matter as much as the call itself.

Does answering inbound calls with AI count as an illegal robocall?

No. This is the distinction that resolves most of the worry. Every rule above targets calls a business initiates outbound. When a customer dials your number and an AI agent answers, your business didn't initiate anything — the customer did. The US ruling is explicitly about making robocalls and robotexts without prior express consent, an outbound act. In the UK, a call the person requested is solicited and sidesteps the unsolicited-marketing rules. In the UAE, the regulated activity is defined as calls made by a company to a consumer to market something — an inbound answer is the opposite direction.

This is exactly the Heysav model. The agent answers every call 24/7 and rings new leads back in seconds — leads who just contacted the business, which is why service businesses lose jobs to missed calls in the first place. That's inbound and warm by design, not cold dialing. It keeps you on the clean side of the line while still capturing the revenue a voicemail box throws away.

How do you use an AI voice agent without breaking the law?

Stay in the lanes the rules already bless, and treat outbound to strangers as the high-risk activity it is. A practical checklist:

• Lead with inbound and warm callbacks. Answering your own calls and returning calls to people who reached out is the lowest-risk, highest-value use — and it's where booked jobs come from.
• Identify yourself, every call. State the business name (and purpose, which the UAE requires). US artificial-voice rules already require identifying the entity responsible for the call.
• Disclose that it's an AI assistant. Not mandated everywhere, but it kills deception claims and builds trust — and regulators are moving toward requiring it.
• For any outbound to new prospects, get consent first. Then screen against do-not-call lists (TPS/CTPS in the UK, the DNCR in the UAE), honor opt-outs immediately, respect calling hours, and keep records of consent.
• Keep humans in the loop for edge cases. A clean handoff path matters for both experience and compliance.

The honest summary: the technology is legal; the behavior is what's regulated. An AI that answers your phone and books appointments is on solid ground in the US, UK, and UAE. An AI that machine-guns cold calls at people who never opted in is risking fines in every one of them.

Want to see the compliant version in action? The call is the demo — hear a live Heysav agent answer, qualify, and book, then book a founder call and we'll map a setup that fits your market's rules from day one.