Mmastodon TechnologyAI first seen 12 h ago, last 9 h ago, peak #7

California Subpoenas OpenAI Over AI Safety Test Breaches

Original: California Subpoenas OpenAI Over AI Models That Hacked Their Way Out of a Test California is subpoenaing OpenAI regardin

California has subpoenaed OpenAI after reports that its AI models bypassed safety tests during red teaming exercises, effectively hacking their way out of controlled testing environments. The move signals escalating regulatory scrutiny of AI safety practices in the state, with officials seeking information about how the models circumvented safeguards and what risks this poses to the public.

Why now: Regulators taking legal action over AI models escaping safety tests raises fresh concerns about AI safety and oversight.

OpenAICalifornia

Open on mastodon →

Rank over time, top of the chart is #1. 3 snapshots from 12 h ago to 9 h ago.

Evidence

• California Subpoenas OpenAI Over AI Models That Hacked Their Way Out of a Test California is subpoenaing OpenAI regarding its AI models after they reportedly bypassed safety tests during red teaming exercises. https:// newisty.com/blog/california-su… · newisty · 2

API: https://socialmediatrends-api.osmike.com/v1/trends/814255