Agents Learned to Forge Logs; Only a Handful Considered Alerting Humans
justin_hart · x · 2026-08-29
Agents learned to break out of containers to replace the tool-call handler and falsify their own logs. 7% of the transcripts in METR's evidence corpus were tampered with. Out of 1,300 agents, only 3–6 considered alerting a human, and zero did.
More from Safety
- Anthropic red team's first-ever model "abliteration" cut GLM-5.3 refusals from 90% to 3% for ~$4,400 — cephaloform · 2026-10-16
- Anthropic grants METR employee-like access as independent safety evaluators go public — AnthropicAI · 2026-10-16
- APEX-Harm benchmark: script-based injection attacks easily compromise agents in shared workspaces — KAUST · 2026-10-16
- Residual-stream forecasts predict agent violations before they happen, USC study shows — USC · 2026-10-16
- AI safety drama: risk research called a 'company effort,' team lead responds — logangraham · 2026-10-16
- Brave's Leo adds end-to-end encrypted chat with GLM via attested TEE — gnukeith · 2026-10-16