Buck Shlegeris

Chronological feed of everything captured from Buck Shlegeris.

paper / buck-shlegeris / May 10 / failed

Evaluating Control Protocols for Untrusted AI Agents

paper / buck-shlegeris / May 10 / failed

LinuxArena: A Control Setting for AI Agents in Live Production Software Environments

paper / buck-shlegeris / May 10 / failed

Auditing Sabotage Bench: A Benchmark for Detecting and Fixing Research Sabotage in ML Codebases