Short answer: in most cases, yes.
Since 2 August 2026, Article 50(1) of the EU AI Act has required AI systems that talk directly with people to make clear that the person is dealing with an AI. That covers the chatbot on your website, the voice agent answering your phones, and an AI agent that emails or books things on your behalf. It applies to businesses outside the EU too, if their AI talks to people in the EU.
A lot of the coverage this year has been about the AI Act being "delayed", which has left some business owners thinking nothing applies yet. That is not the case for this rule. Here is what it says, who it covers, and what a compliant disclosure looks like in practice.
What the rule says
Article 50(1) puts the obligation on providers: AI systems "intended to interact directly with natural persons" must be designed so that those people are told they are interacting with an AI system, unless that is obvious from the point of view of a reasonably well-informed person.
Article 50(5) adds two conditions. The information has to be "clear and distinguishable", and it has to be given "at the latest at the time of the first interaction". It also has to meet accessibility requirements.
On 20 July 2026 the European Commission published guidelines on Article 50, which fill in a lot of the detail. Their list of systems in scope includes voice assistants, chatbots in customer support, e-commerce, finance, healthcare and education, AI hotlines, AI bots on social networks, and AI agents that interact with people while doing a task.
"We're not in the EU. Does this apply to us?"
Quite possibly. Under Article 2, the AI Act applies to providers placing AI systems on the EU market wherever they are based, and to providers and deployers outside the EU "where the output produced by the AI system is used in the Union". The Commission's guidelines repeat that a provider is responsible "regardless of whether the provider is established or located in the Union or in a third country."
So an Indian, UK or US company whose website chatbot answers customers in Germany, or whose voice agent calls patients in Ireland, should treat the rule as applying to those conversations.
Are you the "provider"?
Because Article 50(1) is a provider obligation, this matters. Article 3(3) defines a provider as someone who develops an AI system, or has one developed, and puts it into service under its own name or trademark.
The guidelines give two examples that cover most small businesses:
- A company that develops a chatbot in-house and puts it into service under its own name is the provider.
- A company that takes an existing AI system, modifies it and puts it into service under its own name becomes the provider of the new system.
If you had a custom chatbot or voice agent built and it runs under your brand, you should assume you are the provider. If you simply switched on a vendor's standard chat widget, the vendor is more likely the provider, but it is still your brand in the conversation, so check that the disclosure is switched on and worded properly.
Didn't the Digital Omnibus delay all this?
It delayed some of it, not this part.
The AI "Digital Omnibus" was adopted by the European Parliament on 16 June 2026 and the Council on 29 June, and published in the Official Journal on 24 July 2026. Its headline change was pushing back the rules for high-risk AI systems: to 2 December 2027 for the Annex III use cases and 2 August 2028 for AI built into regulated products.
For Article 50, the only relief is a short grace period for machine-readable marking of AI-generated content under Article 50(2). Generative systems already on the market before 2 August 2026 have until 2 December 2026 to add it. Nothing in the Omnibus moves the date for telling people they are talking to an AI.
What counts as a proper disclosure
This is where the Commission's guidelines are most useful, because they give concrete examples.
For text chatbots, the guidelines suggest plain-language labels or banners (their example is "You are interacting with an AI system"), a first-turn greeting that says it is AI, and where appropriate a persistent label visible throughout the chat, placed near the input box.
For voice and phone agents, they call for an explicit spoken statement at the beginning of the interaction (their example is "This is an AI-powered assistant"), with periodic reminders in longer calls, particularly after an interruption or when the AI's role changes. A beep or tone can support this but is "not considered sufficient" on its own.
For AI agents acting for you, such as an agent that books appointments, manages correspondence or negotiates on someone's behalf, the guidelines say the agent should disclose both that it is AI and on whose behalf it is acting. One of their examples is an email written by an AI agent that carries an AI label at the top.
The guidelines also list approaches that are not enough on their own:
- A disclosure that only appears in your terms and conditions, privacy policy or documentation
- Metadata or watermarks the user cannot see during the conversation
- Vague wording such as calling it an "assistant" without saying it is AI, or a human-like persona that could mislead
- A site-wide line such as "Services on this website use AI"
- Technical descriptions such as "this system uses LLMs" that do not tell the user they are talking to a machine
