Free template · Art. 50(4)

Free AI-Generated Public-Interest Text Disclosure Template

Last updated: 8 November 2026

A public-interest text disclosure template gives deployers ready wording for AI-generated news, current-affairs and similar text under Article 50(4) of the EU AI Act. None is needed if the text had human review or editorial control and a person holds editorial responsibility. The free Article50.io scan loads one public page in a real browser and shows the most severe finding.

Free disclosure templates for AI-generated news, current-affairs and other public-interest text — the second half of Article 50(4) of the EU AI Act, separate from its deep fake rule and with its own editorial review exception.

Scan for free

The templates

Fully AI-generated article · above the headline or byline

<p class="ai-text-notice">
  This article is AI-generated. It was produced by an AI system
  and published without human editorial review.
</p>

Partly AI-generated article

Parts of this article were created using AI, including [the summary / the data section / the translation].

AI-generated summary box

AI-generated summary · This summary was produced automatically from [source] and has not been checked by an editor.

Optional note for editor-reviewed content (the exempt case)

Drafted with AI assistance. Reviewed and edited by [Editor name], who is responsible for this publication.

The last template is optional: it's for content that falls under the editorial review exception, where no disclosure is required but you want to be open about AI assistance anyway.

What Article 50(4) requires for text

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: no disclosure is needed where the AI-generated content has undergone human review or editorial control and a natural or legal person holds editorial responsibility for its publication. An auto-published AI news feed needs the disclosure; an article drafted with AI and signed off by an accountable editor may not.

This is distinct from the deep fake half of the same sub-clause, which covers images, audio and video and has no editorial exception — see the deepfake disclosure template. For how 50(4) and the 50(2) marking rule divide up AI text, read does Article 50 apply to AI-generated text?

Placement

✔Above the headline or next to the byline — before the reader starts the text
✔On AI-generated summaries, at the top of the summary box
✔In syndicated feeds and newsletters, carried with the text rather than left on the original page

Frequently Asked Questions

What counts as public-interest text?

Article 50(4) covers AI-generated or manipulated text published with the purpose of informing the public on matters of public interest — news, current affairs, politics, public health and similar. The Act doesn't define the phrase further; ordinary marketing copy usually falls outside it.

Do I need a disclosure if an editor reviews the AI draft?

Not under Article 50(4), if 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 need to be met.

Is this the same rule as deep fakes?

Same sub-clause, different rule. The first half of Article 50(4) covers deep fake images, audio and video; the second half covers public-interest text, and only this half has the editorial review exception.

Want wording matched to your site?

This is a general template. Get language matched exactly to your site's specific findings — run a free scan.

Article50.io is an automated scanner for the Article 50 transparency rules of the EU AI Act. It scans a website for potential Article 50 transparency gaps and generates a report of what the scan found and what it could not check, with remediation guidance, implementation instructions and disclosure wording you can paste in.

The free scan shows your single most severe finding. Scans are capped at 45 seconds — no signup, public pages only.