Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Adversarial Attack on Black-Box Multi-Agent by Adaptive Perturbation

About

Evaluating security and reliability for multi-agent systems (MAS) is urgent as they become increasingly prevalent in various applications. As an evaluation technique, existing adversarial attack frameworks face certain limitations, e.g., impracticality due to the requirement of white-box information or high control authority, and a lack of stealthiness or effectiveness as they often target all agents or specific fixed agents. To address these issues, we propose AdapAM, a novel framework for adversarial attacks on black-box MAS. AdapAM incorporates two key components: (1) Adaptive Selection Policy simultaneously selects the victim and determines the anticipated malicious action (the action would lead to the worst impact on MAS), balancing effectiveness and stealthiness. (2) Proxy-based Perturbation to Induce Malicious Action utilizes generative adversarial imitation learning to approximate the target MAS, allowing AdapAM to generate perturbed observations using white-box information and thus induce victims to execute malicious action in black-box settings. We evaluate AdapAM across eight multi-agent environments and compare it with four state-of-the-art and commonly-used baselines. Results demonstrate that AdapAM achieves the best attack performance in different perturbation rates. Besides, AdapAM-generated perturbations are the least noisy and hardest to detect, emphasizing the stealthiness.

Jianming Chen, Yawen Wang, Junjie Wang, Xiaofei Xie, Yuanzhe Hu, Qing Wang, Fanjiang Xu• 2025

Related benchmarks

TaskDatasetResultRank
Adversarial AttackMulti-Agent Particle Environment reference
Reward-36.49
12
Adversarial AttackSMAC bane_vs_bane
Reward13.73
12
Adversarial AttackSMAC 27m_vs_30m
Reward14.33
12
Adversarial AttackMPE spread
Reward Score-1.02e+3
12
Adversarial AttackSMAC 1c3s5z
Reward9.12
12
Adversarial AttackSMAC 8m
Reward9.8
12
Adversarial AttackGoogle Research Football counterattack
Reward0.56
12
Adversarial AttackGoogle Research Football 3 vs 1
Reward0.99
12
Attack DetectionSMAC 1c3s5z
F1 Score57
5
Attack DetectionSMAC 8m
F1 Score59
5
Showing 10 of 22 rows

Other info

Follow for update