AI & ML · Aug 10
OpenAI Pauses Astra: Math Skill Still Has to Pass Cyber Gates
Astra may be impressive in capability tests, but OpenAI’s cyber threshold turned evaluation results into a shipping brake.
- Treat cyber capability tests as release gates, not paperwork after the demo.
- Separate model skill from deployment permission, especially for agentic coding tools.
- Watch how labs adjust access, safeguards, and evaluations after Critical findings.