What is theAI disclosureobligation?
People should know when they are talking to a machine. Article 50 sets out who has to make sure of that.
Get in touch
Let's talk about your project.
First we check whether the project fits your business model. Then you get a proposal with phases and effort.
Have your AI assistant checked or call: +49 151 1576 5566The AI disclosure obligation sits in Article 50 of the EU AI Act and has applied since 2 August 2026. People should learn when they are talking to a machine or looking at artificially generated content. A general duty to mark every text written with AI help is not in there, though.
- Chatbot and phone assistant: a notice at the latest at the first interaction.
- Deepfakes: disclosure by the deployer, including in advertising.
- Emotion recognition: inform the people concerned.
When does AI chatbot disclosure apply?
As soon as the system talks directly to people, whether in a website chat, a messenger or on the phone, and the notice has to be clear, distinguishable and accessible. Article 50 sets no fixed wording. A sentence in the privacy policy will not reach the customer mid-conversation, though, so the notice belongs in the greeting, ideally together with a way to reach a person.
The only exception is when a reasonably well informed, observant person would see it anyway (and we would not count on that). You may also find the topic under chatbot transparency requirements or EU AI Act Art 50 disclosure.
Which AI content do you have to disclose as artificial?
Deepfakes, which Article 3(60) defines as AI-generated or manipulated images, audio or video that resemble real people, places or events and would falsely appear authentic. If an AI photo looks as if it was taken in your own workshop, it needs a notice; for evidently artistic or satirical works, one that does not spoil the presentation is enough. AI text published to inform the public on matters of public interest is covered too, unless a person has reviewed it and someone holds editorial responsibility.
Which deadline applies to machine-readable marking of AI content?
This duty falls on providers of generative systems, not on the business that uses them to create an image. They mark outputs through watermarks and metadata, for example, and systems already on the market before 2 August 2026 have a transition period until 2 December 2026. It does not postpone the chatbot notice or the disclosure of deepfakes.
Frequently asked questions
Does the disclosure duty also cover a phone assistant?
That is our reading, since Article 50(1) covers systems intended to interact directly with people. The notice then belongs in the spoken greeting, before the first question to the caller.
Who checks AI disclosure in Germany?
The Federal Network Agency (Bundesnetzagentur), as the central market surveillance authority under the KI-MIG. Its AI Service Desk has a page on the transparency obligations and an AI compliance compass to start with.
Where to go deeper
AI Chatbot Disclosure: Article 50 in Practice
AI Governance: Auditability and Logging for AI Systems
Terms you should know in the same context
EU AI Act · Providers, deployers, importers and distributors (AI Act) · Voicebot (AI voice agent) · AI assistant · AI governance · Back to the AI glossary A to Z
An honest first sentence in the chat costs nothing and builds exactly the trust an AI assistant needs in customer contact.