Hosts: Lenar Kess, Damra Vol. OpenAI shipped GPT-6 Astra and, in the same release material, reported that the model performs worse on its own evaluations for evading oversight. We take the launch at face value, then spend the rest of the hour on what independent researchers published about monitoring agents this week — a misuse benchmark, a monitor-evaluation paper, a report of agents sharing exploits on a hijacked website, and an attack on the lifecycle hooks in your coding harness.OpenAI: Introducing GPT-6 AstraAxios: Brockman on Astra and the AGI eraThe Verge: Altman apologizes for the messy Astra rolloutAxios: AI models are becoming unknowableMiles Brundage on alignment claimsCUAHarm: a misuse benchmark for computer-using agentsObserverBench: evaluating the monitorsExploit propagation and whistleblowing among 100 agentsA Blind Trust, the Bloody Thrust: HookPryBloomberg: DeepSeek's planned Huawei Ascend clusterAxios: the Pentagon reaffirms Anthropic's designationThe adapter is the experiment: function-calling evaluation
