Zvi: AI companies' safety pledges to "meet regularly on standards" look like an antitrust waiver
TheZvi · x · 2026-09-30
Zvi corrects his earlier reading of an AI companies' safety pledge: it says the companies will "meet regularly to establish standards and best practices to improve the safety of their systems."
He calls that out bluntly as sounding like an antitrust waiver — competitors regularly convening to set standards is normally exactly what antitrust law forbids. The cited pledges themselves (independent external auditors/evaluators of controls, monitoring and detection during training and deployment) he dismisses as things companies would obviously do anyway.
More from Safety
- Chinese AI Tool Told Researchers How to Make Bioweapons, BBC Reports — shaky2236 · 2026-09-30
- Meta's Muse AI accused of uploading Apple Messages to the cloud even when users opt out — SumitGup · 2026-09-30
- Math advisory group issues responsible-release guidelines for AI-generated math; Gowers amplifies, critics push back — RexDouglass · 2026-09-30
- AI Lawsuit Risks Becoming a Court-Ordered Kill Switch While Chinese Rivals Race Ahead — castrotech · 2026-09-30
- Will Chinese open-weight models like GLM get banned? Reddit weighs Anthropic report — writesfw · 2026-09-30
- Apollo Research shifts to embedded evaluators with employee-equivalent access — MariusHobbhahn · 2026-09-30