SearchAbout
#AI SafetyStory

MemSyco-Bench: Benchmarking Sycophancy in Agent Memory

MemSyco-Bench is a new benchmark designed to evaluate memory-induced sycophancy in agent systems, addressing the issue where retrieved memories cause agents to over-align with users at the expense of factual accuracy. It includes five tasks to assess an agent's ability to handle memory appropriately.

#AI Safety#Benchmarks#memory-augmented-ai#research#sycophancy
Hugging Face Daily Papers3 min read5d ago
MemSyco-Bench: Benchmarking Sycophancy in Agent Memory
Read original ↗Sign in to upvoteSign in to save
ShareXLinkedInHacker News

Related stories

AgenticSTS: A Bounded-Memory Testbed for Long-Horizon LLM AgentsHugging Face Daily Papers4d agoScarfBench: Benchmarking AI Agents for Enterprise Java Framework MigrationHugging Face Blog5d agoIntroducing GeneBench-ProOpenAI Blog6d agoAutoMem: Automated Learning of Memory as a Cognitive SkillHugging Face Daily Papers5d ago