The European Commission's Non-Binding Guidelines Say Cursory Approval Without Substantive Engagement Cannot Meet the AI Act's Conditions for Human Review or Editorial Control of AI-Generated Text on Matters of Public Interest

The European Commission's guidelines on the transparency obligations in Article 50 of the AI Act, C(2026) 5054 final, are dated July 20, 2026. They say of themselves: "These Guidelines are non-binding. Any authoritative interpretation of the AI Act may ultimately only be given by the Court of Justice of the European Union." According to the guidelines, Article 50 will apply as from August 2, 2026, and texts on matters of public interest that were generated or manipulated by AI and published before that date do not need to be labeled retroactively. One of its duties concerns AI-generated or manipulated text published to inform the public on matters of public interest, and the guidelines set out when it applies and when an exception for human review does. Parts of those pages follow, not the whole.

Which Text the Duty Covers

Under the second subparagraph of Article 50(4), deployers of AI systems that generate or manipulate 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. Published, in the guidelines' reading, means accessible "by an indeterminate, fairly large number of unrelated, potential readers simultaneously and/or successively, whether or not against payment." Matters of public interest, the guidelines say, cover topics including "any economic, financial, political, scientific, or cultural development that may be relevant subject of public debate."

The guidelines give examples on each side. The texts that must be labeled unless they are subject to human review or editorial control include "AI-manipulated corporate reports published on a listed company's website containing investor information." The texts outside the duty include "AI-manipulated text by a consultant for a client advice regarding measures to be taken for regulatory compliance with applicable legislation."

What Counts as Human Review

The exception has two conditions, and both must be met. The text "must have undergone human review or editorial control," and "a legal or natural person must hold editorial responsibility for the publication."

The guidelines define the first condition closely. "Human review refers to the deliberate examination of the substance of the content by one or more natural persons possessing relevant knowledge and professional judgement pertaining to the subject matter under scrutiny (e.g. academic peer review or professional validation chains)." It adds that "Fact-checking the accuracy of the content is a minimum requirement that should be part of that review." Editorial control is control exercised in practice by a responsible editorial entity, such as an editor-in-chief, with the authority to approve, alter or reject the substance of the text on substantive grounds.

What does not count is listed too. "Superficial, solely formal or procedural checks (e.g. spell-checking or grammatical correction), the mere existence of an editorial policy, automated review processes or cursory editorial approval without substantive engagement by the human reviewer or the editorial entity, cannot fulfil the conditions for human review or editorial control for the purposes of this exception." Where AI systems are used to modify, supplement or reformulate content following editorial sign-off, the guidelines say, the resulting content must be treated as AI-generated or manipulated. "Any substantive AI intervention occurring after the human review or editorial control process has taken place will therefore cause the exception to become void." Deployers may adhere to a code of practice, assessed as adequate under Article 50(7), to demonstrate that they meet the condition of human review or editorial control. The guidelines ask what the reviewer did, not whether an editorial policy exists. A sign-off without substantive engagement does not count, and a substantive change by AI after the sign-off undoes it.

Editorial Responsibility

The second condition is responsibility. The person or organization holding it "must hold the ultimate legal responsibility over the publication of the content, including the human review or editorial control." To ensure public accountability and trust, the guidelines say, the identity and contact details of whoever holds editorial responsibility should be made publicly available in an easily findable place, if not yet otherwise available, such as a website's terms and conditions.

The guidelines' examples that meet both conditions include "AI-generated sustainability reports published on a listed company's website having undergone human review by professionals in relevant functions (e.g. compliance)." Those that do not include "AI-generated articles that are reviewed and edited by another AI system and where a human editor performs a mere superficial, grammatical check before publication." For a listed company that publishes reports of this kind drafted with AI, the record of who reviewed them, and what they checked, is what this exception would rest on.