Machine learning (ML) models are susceptible to various security, privacy, and fairness risks. Adversaries with different characteristics (i.e., objectives, knowledge, and capabilities) can collude by executing one attack to amplify others. Existing work lacks a systematic framework to explore collusion among adversaries, and to study the implications of the adversaries' characteristics. We present a framework covering collusion (a) between train- and inference-time adversaries, and (b) among inference-time adversaries. Our framework accounts for factors enabling collusion between adversaries. We propose a guideline to conjecture about the potential for collusion using enabling factors. We use it to explain prior work, conjecture about unexplored collusions, and empirically validate five such cases. Finally, we discuss how adversaries' characteristics influence the potential for collusion.

翻译：机器学习（ML）模型易遭受各类安全、隐私及公平性风险。具有不同特征（即目标、知识和能力）的对手可通过联合执行某项攻击以放大其他攻击效果，实现合谋。现有研究缺乏系统性框架来探索对手间的合谋行为，以及研究对手特征的影响。我们提出一个涵盖（a）训练时与推理时对手之间，以及（b）推理时对手群体内部的合谋框架。该框架考虑了促成对手间合谋的多种因素。我们提出一项准则，利用促成因素推测合谋的可能性，并据此解释已有研究、推测尚未探索的合谋情形，同时通过实验验证了五种此类案例。最后，我们讨论了对手特征如何影响合谋的潜力。