EnvTrustBench is a new agentic benchmark that measures evidence-grounding defects where LLM agents overtrust faulty environmental observations and take incorrect actions.
Agentdojo: A dynamic environment to evaluate prompt injection attacks and defenses for LLM agents.Advances in Neural Information Processing Systems, 37:82895–82920
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
When Agents Overtrust Environmental Evidence: An Extensible Agentic Framework for Benchmarking Evidence-Grounding Defects in LLM Agents
EnvTrustBench is a new agentic benchmark that measures evidence-grounding defects where LLM agents overtrust faulty environmental observations and take incorrect actions.