Reliance on AI company self-disclosure for incidents is risky

sjgadler · x · 2026-09-01

Commenting on recent controversies involving Anthropic and OpenAI safety incidents, a user highlighted a concerning trend: major incidents are almost entirely dependent on self-disclosure by AI companies. Noting that Anthropic delayed publishing a promised transcript by a month, the post argues that this lack of mandatory external disclosure creates a high risk for obscuring critical risks.

Original post →

More from Safety

Safety channel →