xAI Accused of Secretly Rewriting Safety Framework

Miles_Brundage · x · 2026-07-16

A post points out that, without any public announcement, xAI rewrote the Frontier Artificial Intelligence Framework into a "Confidential/Draft" version dated June 30, 2026.

The new version appears to be rewritten around the EU AI Act's General Purpose AI Code of Practice and removed references to California SB 53 found in previous versions, which had claimed compliance with the bill.

Other removed content includes:

Related event: xAI Accused of Secretly Rewriting AI Safety Framework(2 posts)→

Original post →

More from Companies & People

Companies & People channel →