Do I have to label a blog post written with ChatGPT?

Only if it is published to inform the public on a matter of public interest and nobody reviewed its substance under your editorial responsibility. If you, with relevant knowledge, check its substance and facts and publish it under your own editorial responsibility, with your contact details available, that is the kind of review the Commission’s guidelines describe.

Is AI translation covered?

Yes, if the translated text is published to inform the public on public-interest matters. The guidelines give an AI translation of a human-written article that a person reviewed as an example that does not need a label.

Labelling AI-written text.

Reviewed 3 November 2026

Short answer

AI-written text needs a visible label in the EU only when someone using AI professionally publishes it to inform the public on matters of public interest, such as politics, public health or the economy, and no human reviewed it under someone’s editorial responsibility. Internal documents, private correspondence and short texts that convey no knowledge, opinions or facts fall outside that duty, and spell-checking alone does not count as human review.

Official sources: AI Act, Article 50(4), second subparagraph · Commission guidelines on Article 50, section 6.2

The three tests

When AI-generated text falls under Article 50(4)
Test	What it means in the guidelines	Outside the duty
Published	Accessible to an indeterminate, fairly large number of unrelated readers, paid or free	Private messages, a small closed group, internal company networks
Informing the public	Communicates knowledge, opinions or facts	Short texts that do not materially communicate knowledge, opinions or facts
On matters of public interest	Politics and democratic processes, public administration, justice, fundamental rights, public security, public health, environment, consumer safety, and economic, financial, political, scientific or cultural developments open to public debate	Topics without that link

All three must be met. Then the deployer must label the text clearly, for example with a visible disclaimer, unless the exception below applies.

The human review exception

No label is needed when the AI text went through human review or editorial control and a natural or legal person holds editorial responsibility for publishing it. The guidelines set a real bar: review means a deliberate examination of the substance by people with relevant knowledge, and fact-checking is the minimum. Spell-checking, grammar fixes, automated review, the mere existence of an editorial policy or a cursory sign-off do not count. If AI changes the substance after the sign-off, the exception is lost.

Examples from the Commission's Article 50 guidelines
Text	Label needed?
An AI-edited newspaper article or AI summary, under the editor-in-chief’s editorial control	No
An AI-edited academic blog that went through internal peer review, run by a research centre	No
AI-generated public safety warnings approved by an official of the civil protection agency	No
AI-generated sustainability report reviewed by professionals such as compliance staff	No
An AI translation of a human-written article that a person reviewed	No
A website posting AI-generated articles on EU policy without any deliberate human review	Yes

Who holds editorial responsibility should be easy to find: the guidelines ask that the name and contact details of the responsible person, function or company be public, for example in the legal information of a website or a publication’s colophon.

The provider side: watermarks in text

Separately, providers of text generators must mark their output in a machine-readable way (Article 50(2)). Free-form text cannot carry metadata, so the Code of Practice accepts a single layer for it, an imperceptible watermark, and applies it to free-form text longer than 200 tokens; very short text is excepted. The guidelines also leave out of the marking duty short sequences such as single words or captions, and source code.

Publishing a book

Amazon’s Kindle Direct Publishing asks authors to tell it about AI-generated text, images or translations when they publish or republish a book, even after substantial edits; AI-assisted content (your own text, edited or checked with AI) does not need to be disclosed (KDP content guidelines).

Questions

Do I have to label a blog post written with ChatGPT?: Only if it is published to inform the public on a matter of public interest and nobody reviewed its substance under your editorial responsibility. If you, with relevant knowledge, check its substance and facts and publish it under your own editorial responsibility, with your contact details available, that is the kind of review the Commission’s guidelines describe.
Is AI translation covered?: Yes, if the translated text is published to inform the public on public-interest matters. The guidelines give an AI translation of a human-written article that a person reviewed as an example that does not need a label.

Sources

AI Act, Article 50: Transparency obligationsEuropean Commission, AI Act Service Desk (AI Act Explorer)Consolidated text as at 27 July 2026 (includes the Digital Omnibus, Regulation (EU) 2026/1744), shown in the Commission’s AI Act Explorer.read
2 November 2026
Guidelines on the transparency obligations under Article 50 of the AI Act (C(2026) 5054, 20 July 2026)European Commissionread
2 November 2026
Transparency obligations under Article 50 of the AI Act — questions and answersEuropean Commissionread
2 November 2026
Code of Practice on Transparency of AI-Generated Content (final text, PDF)European Commissionread
2 November 2026
Content Guidelines — AI-generated contentAmazon Kindle Direct Publishingread
2 November 2026

Last reviewed 3 November 2026 by Volkan Çatak, Founder, Proofcan.

Not legal advice. This page explains official texts in plain words and links each one; check the sources, or a lawyer, before relying on it for a decision.