CAMA: Exploring Collusive Adversarial Attacks in c-MARL

2026-03-20

AI Summary

Current cooperative multi-agent reinforcement learning (c-MARL) systems are vulnerable to single-point perturbations and white-box attacks, and there is a notable lack of research on coordinated adversarial attacks involving multiple malicious agents. This work proposes the first taxonomy of strategy-level collusive adversarial attacks—namely collective, camouflaged, and spy-type—and introduces a unified attack framework, CAMA, which achieves efficient and stealthy coordination through observation fusion and trigger-based control mechanisms. Theoretical analysis validates the framework’s effectiveness in terms of destructiveness, stealthiness, and cost-efficiency. Experiments across four SMAC II maps demonstrate that the proposed approach exhibits synergistic enhancement among colluding agents, maintains high stealth and stability in long-horizon tasks, and significantly improves attack success rates.

Technology Category

• Multiagent Systems: Adversarial Agents
• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Cooperative multi-agent reinforcement learning (c-MARL) has been widely deployed in real-world applications, such as social robots, embodied intelligence, UAV swarms, etc. Nevertheless, many adversarial attacks still exist to threaten various c-MARL systems. At present, the studies mainly focus on single-adversary perturbation attacks and white-box adversarial attacks that manipulate agents' internal observations or actions. To address these limitations, we in this paper attempt to study collusive adversarial attacks through strategically organizing a set of malicious agents into three collusive attack modes: Collective Malicious Agents, Disguised Malicious Agents, and Spied Malicious Agents. Three novelties are involved: i) three collusive adversarial attacks are creatively proposed for the first time, and a unified framework CAMA for policy-level collusive attacks is designed; ii) the attack effectiveness is theoretically analyzed from the perspectives of disruptiveness, stealthiness, and attack cost; and iii) the three collusive adversarial attacks are technically realized through agent's observation information fusion, attack-trigger control. Finally, multi-facet experiments on four SMAC II maps are performed, and experimental results showcase the three collusive attacks have an additive adversarial synergy, strengthening attack outcome while maintaining high stealthiness and stability over long horizons. Our work fills the gap for collusive adversarial learning in c-MARL.

Problem

• collusive adversarial attacks
• cooperative multi-agent reinforcement learning
• malicious agents
• adversarial robustness
• multi-agent systems

Innovation

• collusive adversarial attacks
• cooperative multi-agent reinforcement learning
• policy-level attack
• stealthiness
• adversarial synergy