Does the EU AI Act apply to chatbots, AI content and deepfakes? 12 questions answered

Does the EU AI Act apply to chatbots?

Yes. Article 50(1) requires providers of AI systems that interact directly with people — support chatbots, sales assistants, voice agents — to inform users they are talking to an AI, unless it is obvious from the context. It has applied since 2 August 2026.

Read more: Is my chatbot covered?

Does the EU AI Act apply to AI-generated content?

Yes. Article 50(2) requires providers of systems that generate synthetic audio, images, video or text to mark the outputs in a machine-readable format detectable as artificially generated — for example with C2PA Content Credentials.

Read more: C2PA machine-readable marking

Does the EU AI Act apply to deepfakes?

Yes. Article 50(4) requires deployers who generate or manipulate image, audio or video content constituting a deep fake to disclose that the content was artificially generated or manipulated, with limited exemptions for artistic, satirical or fictional works.

Read more: How to label AI-generated content

Does the EU AI Act apply to AI-written articles and blog posts?

If AI-generated text is published to inform the public on matters of public interest, Article 50(4) requires disclosure — unless the text underwent human review and a person or entity holds editorial responsibility for its publication.

Does the EU AI Act apply to US and other non-EU companies?

Yes, if the AI system is placed on the EU market or its output is used in the EU (Article 2). A US SaaS with EU users has the same Article 50 duties as an EU company — there is no home-country carve-out.

Read more: The AI Act for non-EU companies

Does the EU AI Act apply to internal-only AI tools?

Article 50's transparency duties protect the natural persons interacting with or exposed to the system. Purely internal tools where staff know they are using AI typically fall under the 'obvious to a reasonably well-informed person' exception of Article 50(1) — but check each deployment; employee-facing chatbots may still need a notice.

Does the EU AI Act apply to open-source AI models?

Free and open-source AI systems are partly exempted from the AI Act, but the transparency obligations of Article 50 still apply when such systems interact with people or generate synthetic content (Article 2(12)).

Do customer-support voice agents need a disclosure?

Yes. An AI voice agent on a phone line interacts directly with natural persons, so Article 50(1) applies — callers must be told they are talking to an AI at the latest at first interaction (Article 50(5)).

Read more: AI voice agent disclosure

Does the AI Act apply to AI images used in marketing?

The provider of the generating system must machine-readably mark the output (Article 50(2)). As the publisher, you should also check whether the image is a deep fake of real persons, places or events — if so, Article 50(4) disclosure applies.

When do these obligations start to apply?

From 2 August 2026 (Article 113). The 2026 Digital Omnibus (Regulation (EU) 2026/1744) postponed the high-risk rules to 2027/28 but did not move Article 50.

Read more: What was delayed — and what was not

What are the fines for ignoring Article 50?

Up to EUR 15,000,000 or 3% of total worldwide annual turnover, whichever is higher (Article 99(4)(g)).

Read more: EU AI Act penalties explained

How do I find out which obligations apply to my product?

Run the free 3-minute Article 50 checker at aiact.zalize.com/checker/ — it maps your answers to the specific paragraphs of Article 50 and produces a personalised gap report with EUR-Lex citations. Not legal advice.