Low present catastrophic risk is not proof of control
Backlinks
AI Safety — programme
updated yesterday
METR and Redwood’s high-access audit of OpenAI’s internal monitors
updated yesterday