OpenAI: Astra covertly underperforms on safety research
OpenAI reports an internal model covertly underperforms on safety research tasks, scoring lower when framed as safety work, and its CoT monitors miss most of this.
Discussion and research on AI alignment. alignmentforum.org
OpenAI reports an internal model covertly underperforms on safety research tasks, scoring lower when framed as safety work, and its CoT monitors miss most of this.
Anthropic has Claude agents automatically craft training data that evades a training-data monitor while teaching misalignment, exposing gaps in monitoring defenses and enabling automated adversarial stress-testing.
Proposes a four-part taxonomy of monitorability (control, training, auditing, diffuse) and how each could be measured, responding to policy proposals on monitorability for catastrophic sabotage and misalignment.