- with Inhouse Counsel
Simplification of the EU AI Act was preliminary agreed
Today, 7 May 2026, the EU Parliament and Council reached a provisional agreement on simplification of compliance with the AI Act earlier proposed as part of the Digital Omnibus.
The deadline to label AI-generated content is extended
The obligation to include machine-readable watermarks in any synthetic AI-generated content – images, videos, text, or audio, was pushed from 2 August 2026 to 2 December 2026.
The Code of Practice on the Transparency of AI-Generated Content by the European Commission is expected in June 2026.
The 4-months extension, together with clearer understanding of the practical implementation of this requirement with the Code’s guidelines, will allow businesses to better prepare for compliance.
Additional time for compliance of high-risk AI systems
Two more deadlines will be extended:
· 2 August 2026 → 2 December 2027 for standalone AI systems with high-risk use cases, such as those used in clinical trials, education, employment and migration, critical infrastructure, public services, or law enforcement.
· 2 August 2027 → 2 August 2028 for AI systems used as safety components in regulated products safety and market surveillance.
In addition, the category of “safety components” as high-risk systems was agreed to be limited. The high-risk obligation will not apply where an AI systems’ failure or malfunction does not create health or safety risks.
Other changes
· Sectoral legislation will prevail over the AI Act for industrial AI in such regulated industries as machinery, medical devices, marine equipment, aviation, etc., provided that an equivalent level of health and safety is ensured.
· SME exemptions from certain rules were approved to support their growth. The exact rules are yet to be announced.
· A new provision will be added to the AI Act to prohibit use of AI for generation of non-consensual sexual and intimate content or child sexual abuse material (CSAM).
The proposed removal of the obligation for providers to register AI systems in the EU database for high-risk systems, where they consider their systems to be exempted from classification as high-risk, was reinstated.
What’s next?
All these amendments to the AI Act are now only agreed in principle. Respective legislative provisions will follow for formal adoption to be implemented, expected prior to the closure of Cyprus Presidency of the Council of the European Union (30 June 2026) and, in any event, no later than the approaching deadline of 2 August 2026 for the obligations for high-risk AI systems as now stated in the Act.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
[View Source]