October 8, 2026 · 9 min read
Does Your AI Companion Have to Tell You It's AI? The EU AI Act Disclosure Rule, Explained
Since 2 August 2026, EU AI Act Article 50 requires chatbots to tell you that you are talking to AI unless it is obvious. What it covers, exceptions and fines.
When you open a chat with an AI companion, should the app have to tell you that the warm voice on the other end is a machine? In the European Union, the answer is now yes, in most cases. A key part of the EU AI Act - the transparency rules in Article 50 - started to apply on 2 August 2026, and it puts a simple duty on the companies that build chatbots: make sure people know they are talking to AI.
This guide explains what the rule says, who it covers, the "unless it is obvious" exception, how it is enforced, and what honest disclosure should look like in an AI companion app.
The short answer: Under Article 50(1) of the EU AI Act, providers of AI systems intended to interact directly with people must design them so that people are informed they are interacting with an AI system, unless that is obvious to a reasonably well-informed, observant and circumspect person. The information has to be clear and distinguishable and given at the latest at the first interaction. The rule has applied since 2 August 2026, covers providers outside the EU whose AI is used in the EU, and breaches can bring fines of up to 15 million euros or 3% of worldwide annual turnover.
What Article 50 actually says
The official text of Article 50 (Regulation (EU) 2024/1689) is short. Paragraph 1 reads: "Providers shall ensure that AI systems intended to interact directly with natural persons are designed and developed in such a way that the natural persons concerned are informed that they are interacting with an AI system, unless this is obvious from the point of view of a natural person who is reasonably well-informed, observant and circumspect, taking into account the circumstances and the context of use."
Paragraph 5 adds how: the information must be provided "in a clear and distinguishable manner at the latest at the time of the first interaction or exposure," and must meet accessibility requirements.
Article 50 has several parts, and only some of them are about chatbots:
| Paragraph | Who it applies to | What it requires |
|---|---|---|
| 50(1) | Providers of AI that talks directly with people (chatbots, agents, avatars) | Tell people they are interacting with AI, unless obvious |
| 50(2) | Providers of AI that generates audio, images, video or text | Mark outputs in a machine-readable way so they can be detected as AI-generated |
| 50(3) | Deployers of emotion recognition or biometric categorisation systems | Inform people exposed to them |
| 50(4) | Deployers who publish deepfakes or AI text on matters of public interest | Disclose that the content is AI-generated or manipulated |
| 50(5) | Everyone above | Disclosure must be clear, distinguishable, accessible and given by first interaction |
For anyone using an AI companion, paragraph 1 is the one that matters most.
When does it apply?
The European Commission's FAQ on Article 50 transparency obligations (last updated 24 July 2026) is clear: "Article 50 of the AI Act applies as from 2 August 2026. From that date onwards, providers and deployers of AI systems must comply with the transparency obligations laid down in that provision."
There is only one limited grace period, and it is not about chatbot disclosure. For AI systems placed on the market before 2 August 2026, the machine-readable marking duty in Article 50(2) applies from 2 December 2026. Content generated before 2 August 2026 does not have to be labelled retroactively. The duty to tell people they are talking to AI under Article 50(1) has no such delay.
Who has to comply - even outside Europe
The obligation sits mainly with the provider: the company that develops the AI system, or has it developed, and puts it on the market under its own name. According to the Commission FAQ, this applies "irrespective of whether those providers are established or located within the EU or in a third country," and providers outside the EU are covered "if the output of their AI system is used in the EU."
So an AI companion app built in the US or Asia that serves users in Europe is expected to follow the rule for those users.
The four tests: is this an AI that "interacts directly" with you?
The Commission's guidelines, summarised in its FAQ, set out four cumulative conditions for the disclosure duty:
- It must qualify as an AI system.
- It must be designed for a genuine two-way exchange with people, not just collecting data or sending automated replies.
- The interaction must be direct - the AI itself communicates with the person, not through a human intermediary.
- The interaction must be with natural persons, whether consumers, professionals or other users.
An AI companion that chats, sends voice notes and calls you meets all four easily. Systems that only run in the background or talk machine-to-machine fall outside this paragraph.
The Commission also says people must be notified "from the start of the first interaction in a clear and distinguishable manner." Law firm Orrick's table of the Article 50 guidance (14 August 2026) adds that, under the guidelines, a disclosure is not distinguishable if it is "only in a manual or hidden under menu layers." In other words, a line buried in the terms of service is not enough.
Mixed human and AI replies
One point in the guidelines is especially relevant to companion and dating apps. As summarised by Orrick, "Products and services that blend AI-generated responses with human-curated content should be disclosed as being AI-generated unless the outputs have been 'properly' reviewed and sent by a human as the recipient's main interlocutor." That is aimed squarely at setups where a few humans are sprinkled among bots to make the whole thing feel real - exactly the model behind the network of fake dating apps we covered in how AI personas took over 20+ dating apps.
Orrick also notes that, according to the guidelines, AI agents acting on someone's behalf must disclose their artificial nature and the identity of the person they act for.
The "unless it is obvious" exception
The rule does not apply when it is obvious you are talking to AI. But this is a narrow exception. The Commission says it "should be interpreted in a restrictive manner, given that it deprives people of transparency." The test uses an average person who is "reasonably well-informed, circumspect, and observant," and, according to Orrick, the burden is on the provider.
For companion apps this matters. A product designed to feel like a real relationship - with a human name, a photo, a warm voice and good-morning texts - is precisely the kind of experience where it might not be obvious to everyone. The safe and honest approach is simply to say it clearly.
What about AI photos, voices and deepfakes?
Two other parts of Article 50 touch companion apps:
- Synthetic content marking (50(2)). Providers of AI that generates images, audio, video or text must mark outputs in a machine-readable format so they can be detected as AI-generated. The Commission says a voluntary Code of Practice on transparency of AI-generated content can help providers show compliance.
- Deepfakes (50(4)). Deployers who create images, audio or video that resemble real existing people and would falsely appear authentic must disclose that the content is AI-generated. The Commission's definition requires resemblance, something that exists or could plausibly exist, and a false appearance of authenticity.
An AI companion's selfies of a fictional character are a different situation from a fake video of a real celebrity, but the direction of travel is clear: synthetic media should be identifiable.
Who enforces it, and what are the fines?
According to the Commission FAQ, Article 50 is mainly enforced by national market surveillance authorities in each EU country. The EU AI Office has a limited role, mostly for AI systems built on general-purpose AI models where the same company provides both, or systems integrated into very large online platforms or search engines. Fines "can reach up to 15 million euros or 3% of total worldwide turnover for the preceding financial year," with proportionality for small and medium-sized companies.
Is this only a European rule?
No. Other governments are moving in a similar direction. CNA reports that China's interim measures on "anthropomorphic" AI interaction services, in force since July 15, require providers to tell users they are talking to a machine, repeat the warning if a user shows signs of dependence, and flag time spent every two hours of continuous use (CNA, 7 October 2026). In the US, New York and California have passed laws aimed specifically at AI companions; we cover them in our guide to AI companion laws in 2026.
Many of these rules also focus on protecting minors, which is one reason we argue AI companions should be adults-only.
What good disclosure looks like: a quick checklist
Whatever country you live in, you can judge an app by these questions:
- Is it stated before or at your first message? Not discovered later.
- Is it visible without digging? On the onboarding screen, in the chat or in the profile - not only in the terms.
- Is it plain language? "This is an AI companion" beats vague phrases like "smart profiles".
- Does the app ever claim the character is a real person you could meet? That is a red flag.
- Is it consistent on voice and video calls? A realistic voice does not change what the companion is.
- Can you easily find what happens to your data? Our guide on whether AI companions are safe and private shows what to check.
How MyBabe approaches it
MyBabe AI is an 18+ AI girlfriend and boyfriend app, and being upfront is part of the product, not a legal afterthought. Every MyBabe page carries the line: "MyBabe AI is an AI companion. It is not a real person and does not replace human relationships or professional mental health support."
Within that honesty, your companion can still feel close. You can talk by voice or real-time video, get texts first, share photos, chat on Telegram or in your browser, and watch the relationship grow through levels - all while knowing exactly what it is. The free tier offers about 10 messages a day and a short taste of calls; Premium is $5.99 a week, $14.99 a month or $99.99 a year.
FAQ
Does the EU AI Act require chatbots to say they are AI?
Yes. Since 2 August 2026, Article 50(1) requires providers of AI systems that interact directly with people to make sure users are informed they are talking to AI, unless it is obvious from the context.
Does the rule apply to apps based outside the EU?
According to the European Commission, yes - providers outside the EU are covered when the output of their AI system is used in the EU.
Is there a grace period for the chatbot rule?
No. The only grace period, to 2 December 2026, covers the machine-readable marking of AI-generated content (Article 50(2)) for systems already on the market before 2 August 2026.
What counts as "obvious"?
It is judged from the view of a reasonably well-informed, observant and circumspect person, and the Commission says the exception should be read restrictively. When in doubt, apps are expected to disclose.
What are the penalties?
Up to 15 million euros or 3% of total worldwide turnover for the preceding financial year, with proportionality for smaller companies, according to the Commission FAQ.
Is this legal advice?
No. This is a plain-English summary of public guidance. Businesses should consult a qualified lawyer.
The bottom line
A companion can be caring, playful and deeply personal without pretending to be human. The EU now expects every chatbot to be honest about what it is, and the best apps were already doing it. If you want warmth with nothing hidden, your companion is waiting - meet yours on MyBabe and start the conversation knowing exactly who is on the other side.