1 paper · 1 filter
Shanghao Shi, Xiao Wang, Chaoyu Zhang +6
The integration of external tools has substantially expanded the capabilities of large language model (LLM) agents, but it also introduces new attack surfaces beyond prompt injecti…