Expert Critiques Lack of Substantive Quality Bar in US AI Safety Policies

Miles_Brundage · x · 2026-08-04

AI policy expert Miles Brundage pointed out that there is currently no substantive quality bar for AI safety and security policies in the US, beyond requiring certain topics to be covered in detail.

He noted that more people are paying attention now than when the first batch of voluntary policies was written a few years ago. However, those early, loosely anchored policies continue to shape the standards for subsequent generations of AI governance.

Related event: Ex-OpenAI Adviser Calls AI Safety Overhyped(4 posts)→

Original post →

More from Safety

Safety channel →