Summary
The paper describes accidental mutation, contagion and recovery as mechanisms through which a local deviation could become a collective failure. Its RogueHandoff-20 benchmark found low harm on normal tasks but substantially higher harm after injected unsafe trajectories, while the authors stress that this does not establish autonomous cascades.
Why it matters
The work shifts agent safety analysis from isolated model failures towards communication paths, susceptibility and recovery across multi-agent systems.
Related coverage
- OpenAI introduces a model-misalignment disclosure framework and six case reports
- Google Home opens device control to third-party AI agents
- A time-stamped study identifies policy violations that evade step-level checks