In this letter, we identify critical vulnerabilities in both protocols proposed by Zhao et al. (published in IEEE TIFS, doi: 10.1109/TIFS.2025.3574993), showing that they are susceptible to collusion attacks. In the first protocol, the leaderP1can determine whether its items are held by any honest party by colluding with other participants. In the second protocol,P1can infer whether all honest parties hold a specific item without any collusion, violating the fundamental privacy guarantees of multi-party private set intersection.

Transformer has been successfully used in practical applications due to its powerful advantages. However, users' input is leaked to the model provider during the service. With people's attention to privacy, privacy-preserving Transformer inference is on the demand of such services. Secure protocols for non-linear functions are crucial in privacy-preserving Transformer inference, which are not well studied. Thus, designing practical secure protocols for non-linear functions is hard but significant to model performance. In this work, we propose a frameworkEastto enable efficient and accurate secure Transformer inference. Firstly, we propose a new oblivious piecewise polynomial evaluation algorithm and apply it to the activation functions, which reduces the runtime and communication of GELU by over 1.5× and 2.5×, compared to prior arts. Secondly, the secure protocols for softmax and layer normalization are carefully designed to faithfully maintain the desired functionality. Thirdly, several optimizations are conducted in detail to enhance the overall efficiency. We appliedEastto BERT and the results show that the inference accuracy remains consistent with the plaintext inference without fine-tuning. Compared to Iron, we achieve about 1.8× lower communication within 1.2× lower runtime.