EU delays high-risk AI Act rules by up to 16 months
The European Union has provisionally agreed to delay the implementation of certain rules within its AI Act by up to 16 months. Existing provisions on high-risk AI systems, originally set for August 2026, will now become active in December 2027 for standalone systems and August 2028 for embedded systems, aiming to streamline and harmonize implementation.
Key Takeaways
- High-risk AI system rules were due to take effect on 2 August, but the timeline has been moved back.
- Standalone high-risk systems will now fall under the AI Act in December 2027.
- High-risk systems embedded into products will become active in August 2028.
- Marilena Raouna said the delay reduces recurring administrative costs and supports legal certainty.
- The agreement is part of the EU’s so-called Omnibus VII legislative package.
Why It Matters
The immediate effect is a longer runway before high-risk AI obligations apply, giving firms more time before the EU’s AI Act enforcement kicks in. That matters because the delay covers both standalone systems and AI embedded into products, not just one narrow use case. The broader signal is that Brussels is prioritizing “streamline” and harmonized implementation over the original August 2026 timetable. For StreamingMeme readers, the concrete watchpoint is whether the provisional agreement becomes final and whether the December 2027 and August 2028 dates hold as written.
Read full article at computing.co.uk