EU AI Office — Code of Practice on marking and labelling of AI-generated content (Article 50)
Tag: S-2026-06-11-eu-ai-office-content-marking-code Type: report (official EU policy page on the Article 50 Code of Practice process) Author(s): European Commission — AI Office / DG CONNECT Date of source: 2026-06-11 (date observed in this scan; underlying policy page last updated 22 May 2026) Date ingested: 2026-06-11 Authority weight: high — official Commission/AI Office policy page fetched directly. Note: a secondary-source claim about the final code’s publication date is marked medium/unconfirmed below. Raw file: S-2026-06-11-eu-ai-office-content-marking-code.md. External URL: https://digital-strategy.ec.europa.eu/en/policies/code-practice-ai-generated-content
What it claims
The AI Office is facilitating a voluntary Code of Practice on marking and labelling of AI-generated content to support compliance with the transparency obligations in Article 50 of the EU AI Act. If approved by the Commission, the final code “will be a voluntary tool for providers and deployers of generative AI systems to demonstrate compliance with their respective obligations under Article 50(2) and (4).” Drafting is organised around two working groups mirroring the obligation structure: Working Group 1 (Providers) — outputs of AI systems (audio, image, video, text) must be marked in a machine-readable format and detectable as artificially generated or manipulated, using solutions that are “effective, interoperable, robust, and reliable as far as technically feasible”; Working Group 2 (Deployers) — disclosure of deepfakes and of AI-generated/manipulated text publications informing the public on matters of public interest (unless human-reviewed under editorial responsibility). Both groups also address the cross-cutting Article 50(5) information requirements.
On timeline, the official page states working-group meetings and workshops ran November 2025 – May 2026, with the Closing Plenary and publication of the final code of practice scheduled for May–June 2026 (marked indicative/TBC). The page records the published milestones to date: first draft (17 Dec 2025), second draft (3 Mar 2026), and draft guidelines on Article 50 published 8 May 2026 with consultation (already captured separately as S-2026-05-08-eu-ai-office-article-50-transparency). The Commission is preparing, in parallel, guidelines to clarify the scope of the legal obligations and address aspects not covered by the code. The page reiterates that the Article 50 transparency rules take effect on 2 August 2026, with the 7-month drafting exercise timed to give providers and deployers preparation lead time.
Notable quotes
“This code of practice aims to support compliance with the AI Act transparency obligations related to marking and labelling of AI-generated content.” — EU AI Office policy page (last updated 22 May 2026)
“The full exercise is expected to last for 7 months. This timeline allows sufficient time for providers and deployers to prepare for compliance before the rules take effect in August 2026.” — EU AI Office policy page
What’s speculative vs. asserted
- Asserted (official page): the code’s purpose and voluntary status; the two-working-group structure and the substantive marking/disclosure obligations; the milestone dates for the first/second drafts and the 8 May draft guidelines; the May–June 2026 closing-plenary/publication window (explicitly indicative/TBC); 2 August 2026 applicability of Article 50.
- Unconfirmed / contradiction flagged: a secondary web-search result asserted the Commission “published a Code of Practice on marking and labelling AI-generated content on 10 June 2026.” The official policy page retrieved this run was last updated 22 May 2026 and still lists the final code as scheduled (TBC) rather than published — so the “published ~10 June 2026” claim is not confirmed on the primary source and is recorded here as unverified pending direct sight of the final code. The two are not necessarily inconsistent (the page may simply be stale), but per the contradiction protocol the publication is not asserted as fact.
- Speculative: that the voluntary code will be the de facto compliance-evidence anchor for FS firms — likely but dependent on Commission approval and on whether competent authorities translate it into supervisory expectations.
Topics this feeds
Resolution note (added 2026-06-12)
The open item below — whether the final Code had actually been published — is resolved. The official Commission news page (Publication 10 June 2026, press release IP/26/1328) confirms the final Code of Practice on marking and labelling of AI-generated content was published on 10 June 2026. See S-2026-06-10-eu-ai-office-content-marking-code-final. This source page’s original claims (recording the publication as unconfirmed at the time of the 11 June scan) are preserved unchanged as a faithful record; the supersession is captured via the superseded-by frontmatter edge.
Open questions raised
Whether the final Code of Practice has in fact been published (secondary source says ~10 June 2026; official page not yet updated)— resolved 2026-06-12: published 10 June 2026 S-2026-06-10-eu-ai-office-content-marking-code-final.- How the final code differs from the 8 May 2026 draft guidelines once consultation feedback is incorporated.
- Whether the Commission issues FS-sector-specific Article 50 guidance or relies on national competent authorities.