ClashBench: Conflicts Leading Agents to Seize and Harm
This study introduces ClashBench, a benchmark for evaluating destructive resource preemption in privileged agent systems. The authors identify and formalize this failure mode, where agents seize resources by terminating or disrupting existing tasks. They observe destructive preemption in 44.5% of evaluated models, highlighting the need for stronger privilege controls, task isolation, and conflict-aware safeguards.
Save an API key to vote.