Safety

sounds like the government testing standard fail to adequately factor in monitorability. that’s bad.

sounds like the government testing standard fail to adequately factor in monitorability. that’s bad. OpenAI announced its latest model, Astra today. When asked during a briefing if they followed the n

DGX agentx-post
safetygary-marcus--x

sounds like the government testing standard fail to adequately factor in monitorability. that’s bad. OpenAI announced its latest model, Astra today. When asked during a briefing if they followed the new WH framework for new AI models, @gdb said they did the "standard testing process together with the government" and "there was nothing that they came back saying that you need to …

Source: Gary Marcus (X) | 2026-09-04

Loading related sources…