AI Risk Wire

All days of the daily brief

Daily brief · 2025-01-23

AI safety: new test results and model versions