Do You Have to Tell People They're Talking to an AI?
By VoxPort AI · Published Jun 12, 2026
Short answer: increasingly, yes. If your business uses an AI voice agent to make or take calls, the rules on telling people they are speaking with a machine are tightening, and the EU AI Act now puts this into law for a large part of the market.
It is one of the most common questions we get, and the answers online are often vague or written for the US, where the rules differ. This is a plain-English guide to where things stand for a business operating in the UK and Europe. It is general guidance, not legal advice, but enough to make sensible decisions.
The rule that matters most: EU AI Act, Article 50
The EU AI Act introduced transparency duties for AI systems that interact with people. Article 50 is the relevant part for voice agents. In plain terms: when a person interacts with an AI system, they should be made aware they are dealing with an AI rather than a human, unless that is already obvious from the context. The obligation sits with whoever deploys and provides the system.
For a voice agent on the phone, "obvious from the context" usually does not apply, because a caller hearing a natural-sounding voice will often assume they are talking to a person. That is exactly the situation the rule targets. So if you run an AI voice agent that handles calls with EU residents, the safe starting position is that the AI nature of the call should be disclosed. The Act is being phased in over different timelines, and the transparency duties are worth getting ahead of rather than retrofitting later.
Does this apply to a UK business?
Possibly, even if you are based solely in the UK. The EU AI Act can reach beyond the EU's borders: if the output of your AI system is used by people in the EU, it can apply regardless of where your business sits. So a UK company making or receiving calls involving people in the EU should not assume it is outside scope.
Within the UK itself there is no single statute mirroring Article 50 word for word yet. But UK GDPR's transparency and fairness principles already pull in the same direction, and deceiving a caller into thinking a machine is a human sits uncomfortably with them. Whether through the EU AI Act, UK GDPR, or simply where regulation is heading, a UK or European business using voice AI should plan to disclose rather than hope to avoid it.
"Disclose up front" or "disclose if asked"?
This is the question businesses actually wrestle with, because they worry that announcing an AI at the start of every call will put callers off. There are broadly two approaches:
Disclose up front. The agent makes clear early in the call that it is an AI assistant. This is the most defensible position: unambiguous, with no room for a caller to feel deceived.
Disclose on ask. The agent does not lead with it but answers honestly the moment a caller asks. This feels more natural to some businesses, but it is weaker ground under a rule that expects people to be made aware, not merely to have the truth confirmed if they happen to ask.
Our view is that up-front disclosure is the safer default, and that it costs less commercially than businesses fear. A brief, confident line that the caller is speaking with an AI assistant rarely loses the call. What loses you trust is a caller who later feels misled. The wording and timing can be designed so disclosure is clear without being clunky, so it is a solvable design problem, not a reason to avoid disclosing.
The other half: who is responsible for the data?
Disclosure is one duty; data protection is the other, and they travel together. Every call involves personal data, and under UK and EU GDPR two roles matter: the controller, who decides why and how data is processed, and the processor, who processes it on the controller's behalf.
In most setups, the business deploying the voice agent to handle its own calls is the controller, and any supplier handling the data acts as a processor under a written data processing agreement. If you are using a voice AI supplier with no clear agreement defining these roles, that is a gap worth closing. This is where many US-built subscription tools fall short for European use: the roles, data location, and processing terms are set to suit the platform, not your obligations. A system designed for UK and EU compliance treats the controller/processor split, the data processing agreement, and data residency as first-class concerns rather than afterthoughts.
A practical checklist
If you are using or about to use an AI voice agent, you should be able to answer:
If you cannot answer some of these confidently, you are not unusual, but they are worth resolving before they become a problem.
Where VoxPort fits
We build voice AI for UK and European businesses with these rules built in from the start: disclosure designed to be clear without hurting the call, a documented controller and processor split, and data handling configured for UK and EU compliance rather than inherited from a US platform's defaults.
If you want to be sure you are on the right side of the disclosure and data rules, book a free consultation and we will walk through where you stand.
This article is general guidance and not legal advice. Rules under the EU AI Act and UK GDPR are detailed and evolving, and you should take specific legal advice for your own circumstances.
VoxPort AI Limited · Company Number NI705644 · info@voxport.ai · +44 7588 376834