Source: JD Supra

EU negotiators have reached a provisional political agreement on a package of amendments called the Digital Omnibus on AI, intended to simplify the AI Act and reduce overlapping regulation. Under that provisional agreement, the compliance date for many “high-risk” AI systems would move from 2 August 2026 to 2 December 2027, while high‑risk AI systems used as safety components in regulated products would have a proposed application date of 2 August 2028. Certain obligations for AI-generated content, including machine‑readable marking or watermarking, would be moved to 2 December 2026. The package would also delay national AI regulatory sandboxes to 2 August 2027 and extend some SME exemptions to small mid-cap companies.

The provisional deal would introduce explicit prohibitions on AI systems used to generate child sexual abuse material and non-consensual intimate content, and it seeks to narrow the definition of “safety component” so that AI features that merely assist users or optimise performance would not automatically be treated as high‑risk unless failure creates a health or safety risk. The text clarifies efforts to reduce product‑safety overlap for AI‑enabled regulated products and lists examples of high‑risk uses — recruitment, employee management, education, credit scoring, access to essential services, biometrics, law enforcement, border management, and safety components in products such as medical devices, machinery, toys and lifts. The agreement is not yet final law and still requires formal adoption by both the Council and the European Parliament; EU institutions say they intend to adopt it before 2 August 2026. Businesses should note some AI Act obligations are already in force and should continue preparing under existing deadlines until amendments are formally adopted.