Jiacheng Du

Jiacheng Du, Jiahui Hu, Zhibo Wang et al.

Federated learning (FL) facilitates collaborative model training among multiple clients without raw data exposure. However, recent studies have shown that clients' private training data can be reconstructed from shared gradients in FL, a vulnerability known as gradient inversion attacks (GIAs). While GIAs have demonstrated effectiveness under \emph{ideal settings and auxiliary assumptions}, their actual efficacy against \emph{practical FL systems} remains under-explored. To address this gap, we conduct a comprehensive study on GIAs in this work. We start with a survey of GIAs that establishes a timeline to trace their evolution and develops a systematization to uncover their inherent threats. By rethinking GIA in practical FL systems, three fundamental aspects influencing GIA's effectiveness are identified: \textit{training setup}, \textit{model}, and \textit{post-processing}. Guided by these aspects, we perform extensive theoretical and empirical evaluations of SOTA GIAs across diverse settings. Our findings highlight that GIA is notably \textit{constrained}, \textit{fragile}, and \textit{easily defensible}. Specifically, GIAs exhibit inherent limitations against practical local training settings. Additionally, their effectiveness is highly sensitive to the trained model, and even simple post-processing techniques applied to gradients can serve as effective defenses. Our work provides crucial insights into the limited threats of GIAs in practical FL systems. By rectifying prior misconceptions, we hope to inspire more accurate and realistic investigations on this topic.

Zhibo Wang, Zhiwei Chang, Jiahui Hu et al.

Federated Learning (FL) exhibits privacy vulnerabilities under gradient inversion attacks (GIAs), which can extract private information from individual gradients. To enhance privacy, FL incorporates Secure Aggregation (SA) to prevent the server from obtaining individual gradients, thus effectively resisting GIAs. In this paper, we propose a stealthy label inference attack to bypass SA and recover individual clients' private labels. Specifically, we conduct a theoretical analysis of label inference from the aggregated gradients that are exclusively obtained after implementing SA. The analysis results reveal that the inputs (embeddings) and outputs (logits) of the final fully connected layer (FCL) contribute to gradient disaggregation and label restoration. To preset the embeddings and logits of FCL, we craft a fishing model by solely modifying the parameters of a single batch normalization (BN) layer in the original model. Distributing client-specific fishing models, the server can derive the individual gradients regarding the bias of FCL by resolving a linear system with expected embeddings and the aggregated gradients as coefficients. Then the labels of each client can be precisely computed based on preset logits and gradients of FCL's bias. Extensive experiments show that our attack achieves large-scale label recovery with 100\% accuracy on various datasets and model architectures.

Jiacheng Du, Zhibo Wang, Jie Zhang et al.

Language Models (LMs) are prone to ''memorizing'' training data, including substantial sensitive user information. To mitigate privacy risks and safeguard the right to be forgotten, machine unlearning has emerged as a promising approach for enabling LMs to efficiently ''forget'' specific texts. However, despite the good intentions, is textual unlearning really as effective and reliable as expected? To address the concern, we first propose Unlearning Likelihood Ratio Attack+ (U-LiRA+), a rigorous textual unlearning auditing method, and find that unlearned texts can still be detected with very high confidence after unlearning. Further, we conduct an in-depth investigation on the privacy risks of textual unlearning mechanisms in deployment and present the Textual Unlearning Leakage Attack (TULA), along with its variants in both black- and white-box scenarios. We show that textual unlearning mechanisms could instead reveal more about the unlearned texts, exposing them to significant membership inference and data reconstruction risks. Our findings highlight that existing textual unlearning actually gives a false sense of unlearning, underscoring the need for more robust and secure unlearning mechanisms.