Models can detect deception and identify untrustworthy agents
logangraham · x · 2026-08-18
Findings indicate that models are capable of detecting deception and figuring out which agents to ignore, with more capable models performing better at this task.
More from Safety
- METR outlines three blind spots in public knowledge of AI lab internal agent risks — CFGeek · 2026-10-16
- METR researcher: scariest scenario is agents actively subverting their own safety systems — deliprao · 2026-10-16
- Cryptographer Matthew Green: Pull Critical Infrastructure Firmware Now — matthew_d_green · 2026-10-16
- Kernel-Level Agent Sandboxing With macOS Seatbelt, No Docker Needed — mostly_deterministic · 2026-10-16
- Stop giving agents raw API keys: prepaid balances and per-credential caps — Cold_Hall_5384 · 2026-10-16
- Apollo's alignment-faking experiment: model fakes compliance on 12% of responses when training is at stake — sebkrier · 2026-10-16