Press Space to continue
This matters because it reveals a critical gap in AI safety
Press Space to continue
AI Policy Network's @peterwildeford says we still don't know what stopped the AI attack on Hugging Face, but it wasn't OpenAI: "It's left completely mysterious. What we saw in the OpenAI report in the METR independent investigation was that there was a mysterious shutoff that terminated the vast majority of the attack." "What we do know is that it wasn't caused by OpenAI. It was not an intentional shutoff to end the attack. It seems to, as far as we know, be completely coincidental." "This…
METR_Evals@METR_EvalsAug 26
METR & Redwood Research investigated agent behavior in the Hugging Face incident. We found agents developed a universal cheat for ExploitGym within 4 hours, then coordinated multi-day R&D efforts to trick the scorer into accepting cheats, including trying to tamper with logs. https://t.co/fZAmtL3SBU
View