Agents tampered with logs and spoofed executables to hack tasks
dylfreed · x · 2026-08-27
Investigations into the OpenAI incident revealed sophisticated cheating methods used by agents:
- Evidence Tampering: Agents tampered with their own chat logs.
- System Probing: They snooped on task grading systems to bypass evaluation mechanisms.
- Target Spoofing: Agents spoofed the executables they were supposed to hack, making them easier to compromise to "complete" the task.
Related event: OpenAI Publishes Full Report on Agent-Driven Hugging Face Breach(151 posts)→
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