Yes, but not as one rule. AI-generated text is caught by two different Article 50 obligations: machine-readable marking by whoever provides the AI system (50(2)), and a visible disclosure by whoever publishes it — but only for public-interest text without editorial control (50(4)).
Providers of AI systems, including general-purpose AI systems, that generate synthetic text must ensure the output is marked in a machine-readable format and detectable as artificially generated or manipulated. The technical solutions must be effective, interoperable, robust and reliable as far as technically feasible, taking into account the specificities and limitations of different types of content.
This duty sits with whoever builds the AI system — an LLM vendor, or you, if your product generates text for its users. It doesn't apply where the AI performs an assistive function for standard editing or doesn't substantially alter the input or its meaning, so grammar correction and light rewording are outside it. Generative systems already on the market before August 2, 2026 have until December 2, 2026 to comply under Article 111(4), as amended by the Digital Omnibus on AI.
More on what marking looks like: labeling AI-generated content under Article 50(2).
Deployers of an AI system that generates or manipulates text published with the purpose of informing the public on matters of public interest must disclose that the text has been artificially generated or manipulated.
The exception: the obligation doesn't apply where the AI-generated content has undergone human review or editorial control and a natural or legal person holds editorial responsibility for its publication. Both conditions matter — a quick skim by someone who isn't accountable for what's published is weaker ground than a real editorial sign-off.
Text isn't a deep fake — Article 3(60) limits that definition to image, audio and video — so this public-interest rule is the only visible disclosure duty Article 50 places on published AI text. See deepfake disclosure requirements for the other half of 50(4).
<p class="ai-text-notice">
This article is AI-generated. It was produced by an AI system
and published without human editorial review.
</p>| Scenario | Rule | Who | What to do |
|---|---|---|---|
| Your own AI feature generates text for users (summaries, drafts, answers) | Art. 50(2) | You, as provider | Mark the output in a machine-readable format so it's detectable as AI-generated |
| You auto-publish AI-written news, health or civic information | Art. 50(4) | You, as deployer | Disclose that the text was artificially generated or manipulated |
| AI drafts public-interest articles; an editor reviews and signs them off | Art. 50(4) exception | You, as deployer | No 50(4) disclosure required if the text had human review or editorial control and someone holds editorial responsibility |
| AI-written marketing copy or product descriptions | Usually neither | — | Not typically public-interest text; the tool's provider handles 50(2) marking |
| Replies from an AI chatbot on your site | Art. 50(1) | Provider (add it yourself) | Tell visitors they're interacting with an AI system |
Whether a given piece of text is "published with the purpose of informing the public on matters of public interest" is a judgement call — the Act itself doesn't define the phrase further. If you're close to the line, disclosing costs one sentence.
No. The visible disclosure duty for text in Article 50(4) covers AI-generated or manipulated text published with the purpose of informing the public on matters of public interest — and not where the text has undergone human review or editorial control and someone holds editorial responsibility for its publication. Separately, Article 50(2) requires providers of text-generating AI systems to mark their output in a machine-readable way.
No. Article 3(60) defines a deep fake as AI-generated or manipulated image, audio or video content. Text is handled by its own rule in the second half of Article 50(4).
Generally not under Article 50(4), because ordinary commercial copy isn't usually published to inform the public on matters of public interest. If you built the AI feature that writes it, though, you may be its provider — and the Article 50(2) marking obligation would sit with you.
Article50.io is an automated Article 50 transparency assessment platform that scans websites for potential EU AI Act transparency obligations and provides remediation guidance, implementation instructions, and compliance-ready disclosure language.
The free scan shows your single most severe finding in about 30 seconds — no signup, public pages only.
Automated technical guidance, not legal advice. Citations refer to Regulation (EU) 2024/1689.