MemEvoBench is presented as the first standardized benchmark for long-horizon memory safety in LLM agents, covering adversarial memory injection, noisy tool outputs, and biased feedback across QA and workflow tasks.
Title resolution pending
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CL 2representative citing papers
Agent-SafetyBench shows no tested LLM agent exceeds 60% safety score, attributing failures to lack of robustness and risk awareness.
citing papers explorer
-
MemEvoBench: Benchmarking Safety Risks from Memory Misevolution in LLM Agents
MemEvoBench is presented as the first standardized benchmark for long-horizon memory safety in LLM agents, covering adversarial memory injection, noisy tool outputs, and biased feedback across QA and workflow tasks.
-
Agent-SafetyBench: Evaluating the Safety of LLM Agents
Agent-SafetyBench shows no tested LLM agent exceeds 60% safety score, attributing failures to lack of robustness and risk awareness.