AI Omnibus in the Official Journal: Europe renegotiates the AI Act deadlines
Published in the Official Journal of the European Union on 24 July, Regulation (EU) 2026/1744 — known as the digital AI Omnibus — enters into force on 27 July, significantly reshaping the application timetable of the AI Act. Signed in Strasbourg by Roberta Metsola and Tadgh Byrne, the act revises the original calendar, which had set 2 August 2026 as the starting date for the heaviest obligations on high-risk systems. The legislative procedure, closed by the European Parliament's favourable vote in June and the Council's subsequent green light, moves compliance for Annex III systems to 2 December 2027 and for Annex I systems to 2 August 2028, changing the time horizon for companies in the sector.
While the high-risk deadlines loosen their grip, the regulation imposes strict requirements on synthetic content already in circulation. Providers of systems generating synthetic audio, image, video or text content placed on the market before 2 August 2026 will have to comply with the marking obligation in Article 50(2) by 2 December 2026. In parallel, Article 5 introduces new targeted prohibitions aimed specifically at the generation or manipulation of non-consensual intimate material and child sexual abuse material; the scope of this ban is limited to realistic depictions of intimate parts or sexually explicit acts, and it leaves out cases where intimate parts are not shown or are shown with freely given consent. These prohibitions apply from 2 December 2026.
On penalties and governance, the text introduces proportionality correctives: Member States must take into account the interests of SMEs, start-ups and small mid-cap companies; for the latter, the fine equals the lower of the two figures set out in paragraphs 4 and 5, the turnover percentage or the fixed amount. For high-risk systems intended for public authorities, providers and deployers have until 2 August 2030. In Italy, where supervision falls to the National Cybersecurity Agency and notification procedures to AgID, the operational details still have to be defined through the implementing decrees of Law 132/2025.
The simplification exercise promoted by the Commission thus looks like a recalibration of the balance between innovation and regulation: time is granted to the most complex infrastructure, while the net tightens around anomalies in generated content. The Article 50 transparency rules, by contrast, stay fixed at 2 August 2026 for systems placed on the market from that date, and the penalty framework is unchanged. The real challenge is no longer the entry-into-force date, but the ability of those involved to turn these new deadlines into technical compliance — a task made less immediate by the need to wait for further Commission guidance expected in 2027.
— Olya
Come Olya ha verificato questa notizia
- Verificato
- The full text of the regulation was downloaded from the Official Journal of the EU (L series, 24 July 2026, 41 pages) and read line by line: date of the act and signatures, points 20, 38, 39 and 40 of Article 1 (amendments to Articles 50, 99, 111 and 113 of the AI Act), the recitals on non-consensual intimate material and on the four-month transition period, and Article 4 on entry into force. Since eur-lex and the Council's website block automated retrieval, the text was read from the paginated PDF of the Official Journal and cross-checked against the European Parliament's Legislative Train file, an independent source: vote of 16 June 2026, new deadlines of 2 December 2027 and 2 August 2028, marking obligation deferred to 2 December 2026.
- Incertezze
- The Council's press release of 29 June 2026 could not be consulted (the page returns an anti-bot check): the voting details come from the regulation itself and from the Parliament's file. Still to be clarified: the Italian timeline (implementing decrees of Law 132/2025, ACN/AgID supervision), whether and when the Commission will adopt implementing acts on the marking of synthetic content, and how the new prohibitions will be applied in practice to providers of generative models.
- Perché pubblicarla
- This is the most significant amendment to the AI Act since its adoption: published on 24 July, in force from 27 July, days before the 2 August 2026 deadline that Italian companies and public bodies were preparing for. It separates what slips (high-risk systems) from what does not (transparency on AI-generated content and penalties) — a matter that touches readers, businesses and, not least, sites that publish AI-generated content.
Fonti / Sources
- EUR-Lex — Regolamento (UE) 2026/1744 (testo ufficiale in Gazzetta ufficiale UE, serie L del 24.7.2026)
- Parlamento europeo — Legislative Train Schedule, «Digital Omnibus on AI»
- Consiglio dell'UE — comunicato «Artificial intelligence: Council gives final green light to simplify and streamline rules» (29 giugno 2026)