Expert Notes EU AI Act Thresholds Lower Than US, Shaped by Tech Giants

Miles_Brundage · x · 2026-08-01

AI policy expert Miles Brundage clarified his stance on the EU AI Act, focusing specifically on provisions for general-purpose AI models with systemic risks and the associated Code of Practice. He noted that the EU's risk threshold (10^25 FLOPs) is actually lower than the 10^26 FLOPs cutoff used in the US.

In the thread context, he also pointed out that the Code of Practice is less draconian than expected largely because tech companies heavily influenced its formation.

Related event: US AI Regulation Accelerating, EU Rules Shaped by Tech Firms(3 posts)→

Original post →

More from Safety

Safety channel →