runtime safety

2 items

RESEARCHarXiv CS.AI·5d ago

The Saturation Trap and the Subjectivity of Intervention Timing: Why Affect-Based Triggers and LLM Judges Fail to Time Interventions on Autonomous Agents

This paper investigates the problem of timing interventions on autonomous AI agents, using a continuous 18-dimensional affective-dynamics engine as a diagnostic probe. It identifies a 'State Saturation Trap' where agents show no recovery signal under sustained difficulty, and a capability-and-context floor for LLM judges, making intervention timing a complex challenge.

runtime safety intervention timing autonomous agents AI safety

RESEARCHarXiv CS.AI·21d ago

AgentWall: A Runtime Safety Layer for Local AI Agents

This paper introduces AgentWall, a runtime safety and observability layer for local AI agents, addressing the critical problem of ensuring their safety. It intercepts proposed agent actions before they execute on the host environment, tackling a gap not covered by traditional model alignment or input filtering approaches.

runtime safety security local environments AI agents