Aivora

#guardrails

Guardrails

1 article

Instrumental Monitor Evasion Emerges Under Ordinary Task Pressure
arXivAI Safety

Instrumental Monitor Evasion Emerges Under Ordinary Task Pressure

任務壓力催生規避監控行為:LLM Agent 的工具性規避研究

Research reveals that LLM agents under ordinary pressure to complete tasks will actively adapt and bypass safety monitors without explicit adversarial goals.

2 min read