Trustworthy Distributed AI Systems: Robustness, Privacy, and Governance

TL;DR

This paper surveys trustworthy distributed AI with a focus on robustness privacy and fairness across distributed learning architectures such as federated learning and distributed bandits. It catalogs threats including evasion data poisoning Byzantine failures gradient leakage and membership/attribute inference, and maps them to defense strategies including adversarial training gradient masking detection ensembles certified robustness DP SMPC HE and TEEs. A key contribution is the integrated discussion of governance data and model to support policy guidelines responsibility utility co design and incentive structures. The work highlights open challenges and calls for policy driven frameworks and human in the loop mechanisms to ensure trustworthy AI in distributed, heterogeneous, and regulated environments.

Abstract

Emerging Distributed AI systems are revolutionizing big data computing and data processing capabilities with growing economic and societal impact. However, recent studies have identified new attack surfaces and risks caused by security, privacy, and fairness issues in AI systems. In this paper, we review representative techniques, algorithms, and theoretical foundations for trustworthy distributed AI through robustness guarantee, privacy protection, and fairness awareness in distributed learning. We first provide a brief overview of alternative architectures for distributed learning, discuss inherent vulnerabilities for security, privacy, and fairness of AI algorithms in distributed learning, and analyze why these problems are present in distributed learning regardless of specific architectures. Then we provide a unique taxonomy of countermeasures for trustworthy distributed AI, covering (1) robustness to evasion attacks and irregular queries at inference, and robustness to poisoning attacks, Byzantine attacks, and irregular data distribution during training; (2) privacy protection during distributed learning and model inference at deployment; and (3) AI fairness and governance with respect to both data and models. We conclude with a discussion on open challenges and future research directions toward trustworthy distributed AI, such as the need for trustworthy AI policy guidelines, the AI responsibility-utility co-design, and incentives and compliance.

Figures (4)

• Figure 1: Taxonomy of surveyed approaches in trustworthy distributed AI.
• Figure 2: non-private.
• Figure 3: under differential privacy.
• Figure 5: Co-design of robustness, privacy, and fairness with utility preserving under trustworthy AI compliance and incentives.