Advanced Persistent Threats (APTs), due to their stealthiness and complexity, have become a significant security challenge for modern enterprises, often causing severe economic losses. To address these threats, researchers have proposed using provenance graphs to model system entities and their dependencies, aiming to capture the complex scenarios of APT attacks. However, existing Provenance-based Intrusion Detection Systems (PIDS) still suffer from the following challenges: (1) Historical interaction information loss due to the truncation of long-term interaction scenarios; (2) The difficulty in capturing long-distance dependencies leads to the loss of crucial contextual information; (3) Existing methods struggle to balance detection efficiency and granularity. We introduce PanThreat, an online detection system that performs fine-grained, real-time analysis of host system logs to identify malicious activities. PanThreat combines attributes encoding through Word2Vec and position encoding using Laplacian feature matrices, while retaining long-term interaction histories and effectively modeling long-range dependencies within provenance graphs. This integrated approach significantly enhances detection accuracy. Additionally, PanThreat leverages the parallel processing capabilities of Graph Transformers to improve detection efficiency. Evaluations on the DARPA E3 dataset and StreamSpot database demonstrate PanThreat's effectiveness in detecting complex APT attacks, outperforming four state-of-the-art methods while maintaining an average processing speed of 58,140 events per second.

Machine Learning (ML) based malware classifiers are vulnerable to exploitation during the training phase due to the necessity of regular retraining with samples collected from the wild. Recent studies have highlighted the efficacy of backdoor attacks in the malware domain, where attackers can manipulate the model during training by injecting samples embedded with specific triggers, causing the model to establish an association between the trigger and a designated class, thereby achieving evasion of detection. While research on backdoor attacks has been extensively explored in the field of computer vision, it has been largely overlooked in the malware domain. Unlike in the computer vision domain, the threat model in the malware domain typically restricts attackers to employing clean-label attacks (i.e., attackers do not have control over the labeling of poisoned data). However, clean-label attack methods are generally less effective compared to those that involve embedding triggers and altering sample labels to the target class (called corrupted-label attacks). To address this limitation, we propose a simple yet effective method that involves Poisoning Malware-Similar Benignware (PMSB) instead of random selection, thereby approximating the scenario of corrupted-label attacks and enhancing the effectiveness of clean-label attacks. Additionally, we introduce three similarity measurement methods based on feature-based distance, distribution-based distance, and contribution-based difference to select malware-similar benignware. Comprehensive evaluations across three different trigger types and three datasets demonstrate the superiority and general applicability of PMSB.