作者位次 #6/7ACL 2026Poster· Natural Language Processing
Evo-Attacker: Memory-Augmented Reinforcement Learning for Long-Horizon Tool Attacks on LLM - MAS
Bingyu Yan · Xiaoming Zhang · JinYu Hou · Chaozhuo Li · Ziyi Zhou · Yiming Hei · Litian Zhang
摘要
While Large Language Model-based Multi-Agent Systems (LLM-MAS) demonstrate remarkable capabilities in solving complex tasks by orchestrating specialized agents and external tools, the implicit trust in tool outputs creates a critical attack surface. Existing tool attacks are limited by domain spe...