ASLEval: Measuring Privacy Exposure Displacement in LLM Agent Sessions
A new benchmark quantifies how LLM agents shift privacy exposure across sessions, informing risk assessment of agentic deployments.
A new benchmark quantifies how LLM agents shift privacy exposure across sessions, informing risk assessment of agentic deployments.
Auto-scored: The paper introduces a measurement framework for privacy exposure in LLM agents, a minor capability and evaluation development with partial mitigation value.