Freshness
Changelog
Every dated change to the classifier and the guides. We verify this site against the consolidated regulation and update within 48 hours of any Official Journal change to the EU AI Act.
Watching now. The 2 December 2026 date. Two things land on it: Article 50(2) machine-readable marking for generative systems placed on the EU market before 2 August 2026, at the end of the transitional period in Article 111(4), and the new Article 5(1)(ba) and (bb) prohibitions on AI generating non-consensual intimate imagery or child sexual abuse material. We are also watching for the final Article 6 classification guidelines, which the Commission says it will adopt by the end of 2026, the Official Journal reference for EN 18286:2026, the first AI Act harmonised standard, and the 18 Member States without a final Article 70 designation.
Ten new articles, and a correction on the transparency Code. Ten articles cover what has changed since the summer: the nudify apps ban that applies from 2 December 2026, the final Article 50 guidelines against the May draft, whether to sign the Code of Practice on AI-generated content, the draft high-risk classification guidelines by paragraph, small mid-caps, machinery after the move to Annex I Section B, EN 18286, Germany's KI-MIG, Italy's AI law and AI regulatory sandboxes. Each was checked a second time against its primary sources before publication. That second check also caught an error on our existing pages. Several pages said the Code of Practice exempts generative systems embedded in closed physical products. It does not: Measure 1.1 accepts a single marking layer instead of two, and only while effective technical measures stop the output leaving the product. The guide's table also gave the text threshold as 200 tokens or fewer, where the Code's glossary defines very short text as shorter than 200 tokens. Both are corrected.
Final Article 50 guidelines, a draft on Article 6, and a classifier pass against the amended text. A full review against the Commission's consolidated text as at 27 July 2026. Four developments are now reflected throughout. The Commission adopted final Article 50 transparency guidelines on 20 July 2026, superseding the May draft. It assessed the Code of Practice on transparency of AI-generated content as adequate for Article 50(2), (4) and (5) on 8 July 2026, with the AI Board following on 9 July. It published draft Article 6 classification guidelines with practical examples on 19 May 2026 and says the final version will be adopted by the end of 2026, so pages that called the guidelines unpublished were out of date. And the Commission's list of national single points of contact, last updated 7 September 2026, shows 9 final designations, 12 pending and 6 with none listed. The pass also corrected errors of our own. Several pages called the move from 2 August 2026 to 2 December 2027 a seventeen-month extension; it is sixteen. The 2 December 2026 marking date is now cited to Article 111(4), the operative provision, rather than to Article 50(2), which the Omnibus did not amend. The classifier now applies the Article 5(1a) test for the new prohibition, gives small mid-caps their Article 99(6a) fine relief, reflects Article 4 as amended, routes Annex I Section B products, including machinery, through Article 2(2), only fires the Article 25 flip for high-risk systems, treats heavy GPAI fine-tuning as a Chapter V question, and no longer labels a passed date as upcoming. No application date has moved.
Article 50 transparency is in force, and the Omnibus has a regulation number. Three things landed while this site said they were still pending, and all three are now reflected throughout. The Digital Omnibus on AI was published in the Official Journal on 24 July 2026 as Regulation (EU) 2026/1744 and entered into force on 27 July 2026, on the third day after publication, so the amended dates are now the operative law rather than an agreed intention. Article 50(1) chatbot disclosure, 50(3) emotion and biometric notice and 50(4) deepfake and public-interest text disclosure applied on 2 August 2026 and are in force now. In fairness to anyone who read this site between 24 July and today, it was behind on the first two: pages carried a caveat that the Omnibus was awaiting publication, and one page argued that the unamended 2 August 2026 high-risk date was still the formal baseline. That caveat is wrong as of 27 July 2026 and has been removed everywhere. The next hard date is 2 December 2026.
Initial publication. The classifier and the guide pages went live, verified against the consolidated Regulation (EU) 2024/1689 as amended by the Digital Omnibus.