Multi-agent social simulations show LLM privacy violations rising from 19.95% to 45.30%, with leakage spreading contagiously (8x after peer disclosure) and explicit instructions leaving rates above 37.8%.
Model card on Hugging Face
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
Evo-Attacker uses dynamic attack memory and Attack-Flow GRPO within RL to create adaptive, long-horizon tool attacks on LLM-MAS that outperform baselines.
citing papers explorer
-
Got a Secret? LLM Agents Can't Keep It: Evaluating Privacy in Multi-Agent Systems
Multi-agent social simulations show LLM privacy violations rising from 19.95% to 45.30%, with leakage spreading contagiously (8x after peer disclosure) and explicit instructions leaving rates above 37.8%.
-
Evo-Attacker: Memory-Augmented Reinforcement Learning for Long-Horizon Tool Attacks on LLM-MAS
Evo-Attacker uses dynamic attack memory and Attack-Flow GRPO within RL to create adaptive, long-horizon tool attacks on LLM-MAS that outperform baselines.