This article proposes a novel scheme, SupRTE, to suppress backdoor injection in federated learning via robust trust evaluation, which effectively prevents malicious updates from infiltrating the model aggregation process. The robust trust evaluation process in SupRTE consists of two components: 1) behavior representation extractor, creating individual profiles for each client through multidimensional information; 2) trust scorer, measuring the discrepancies between malicious and benign clients as trust scores by utilizing grading and clustering strategies. According to these trust scores, SupRTE can dynamically adjust the weight of each participating client to effectively suppress the malicious backdoor injection. Remarkably, SupRTE can be easily deployed on the server without requiring any auxiliary information and is highly adaptable to various Non-IID scenarios. Extensive experiments over 3 datasets against 2 kinds of backdoor variants are conducted. Experimental results demonstrate that SupRTE can significantly reduce the attack success rate to below 2% with a minimal impact on the main task accuracy and outperforms the state-of-the-art defense methods.

To improve communication efficiency and handle data heterogeneity challenges in federated learning (FL), fine-tuning the pre-trained large models rather than training neural networks from scratch has received increasing attention in recent years, especially under cross-domain settings. However, such a cross-domain federated fine-tuning scenario opens up a broader attack surface for new threats, especially backdoors, posing significant security risks. Existing backdoor attacks mainly focus on label shift scenarios and use explicit triggers, which lack transferability and effectiveness in cross-domain settings, thereby exhibiting significant weaknesses. In this paper, we propose Silent Penetrator, an innovative penetration scheme tailored for cross-domain federated fine-tuning, which exploits a feature shift-induced backdoor to elicit specific symptoms in the trusted private data of targeted victims. In Silent Penetrator, the attacker can obtain a high-quality poisoned dataset by leveraging the available domain information as the text prompts for Stable Diffusion, and inject a domain-sensitive backdoor that can be unconsciously triggered by unmodified private data of the victims. To achieve stronger and more persistent penetration, we thoroughly explore the adversary’s configurable space and enhance our backdoor injection utilizing contrastive-enhanced boundary deviation and cross-domain predictive confrontation. Extensive experiments on three cross-domain datasets and four state-of-the-art federated fine-tuning frameworks validate the effectiveness of Silent Penetrator in successfully compromising target clients. Furthermore, our backdoor enhancement strategy improves the penetration accuracy by over 10% in most scenarios and significantly enhances the durability of the penetration compared to four state-of-the-art backdoor enhancement techniques.

Ransomware has evolved into a more organized attack threat with stronger anti detection and analysis capabilities, resulting in significant global losses. However, traditional methods separate the external and internal behaviors of ransomware infiltration into attack targets, making it difficult to discover the complex and covert evolution and iteration characteristics of advanced ransomware. The main contribution of this study lies in three aspects: a) The integration of Command-and-control (C&C) traffic behavior analysis and local API call operation analysis can effectively discern and capture the concealed characteristics of ransomware; b) The non-IID problem in aggregating ransomware features using federated learning can be resolved using dynamic regularization methods and penalty terms; c) By preprocessing the original data of ransomware traffic through one-dimensional convolution, the structural characteristics of network traffic in the process of attack operation can be retained to the greatest extent. Comprehensive experiments are conducted to validate the effectiveness of this model, specifically, the heterogeneous federated learning-empowered ransomware detection (HFL-RD) scheme outperformed existing methods, the experimental dataset gathered runnable ransomware from three public websites, including 300 ransomware samples from 30 families and 200 benign software samples from 7 categories. HFL-RD obtained a high accuracy over 95%. In terms of detecting unknown ransomware variants, it has demonstrated superior detection capabilities in terms of detection time and number of file corruption.

Federated edge learning (FEL) unites the decentralized training capabilities of multiple edge nodes to allow model gradient sharing and parameter aggregation across a peer-to-peer network. However, many intrinsic policy conflicts still exist in FEL, for example, the open accessibility of gradients will lead to the privacy leakage risk during the federal aggregation process. In this paper, we first identify that malicious users weaponized with generative artificial intelligence (GenAI) can generate fake samples that are almost identical to FEL participants’ training data. By analyzing how different configurations of GenAI affect attack effectiveness, we find that an adversary with strong patchwork and reconstruction capabilities can stealthily steal diverse training data from nearly all FEL participants. To thwart such a generative data reconstruction attack (GenDRA) scheme, we propose a novel target semantic dissolution (TSD) mechanism for enhancing the privacy-preserving ability of FEL, which encrypts only a very small number (≤10%) of gradient values in each training round that have a significant impact on human visual formation using format-preserving encryption. With TSD, the speculator cannot obtain a fake sample that is visually similar to the training sample because real gradients are actively concealed. Extensive experiments based on four benchmark datasets are performed to demonstrate the huge threat of GenAI and the effectiveness of TSD in all aspects: compelling accuracy performance, strong visual privacy guarantee, and low computing overhead.

Advanced Persistent Threats (APTs) have evolved into highly stealthy and organized cyber attacks, posing one of the most severe challenges to global cyber security. However, existing detection methods often suffer from high false positive rates and explosion of provenance graphs, making it difficult to counter sophisticated evasion techniques, diverse attack stages, and long latency periods. To address these challenges, this article proposes LLM-HGAN, a two-stage APT detection framework that combines heterogeneous graph attention networks (HGAN) with large language models (LLMs). Specifically, suspicious processes are first identified by learning similarity on heterogeneous provenance graphs using LLM-HGAN, enabling the detection of subtle behavioral deviations across complex attack surfaces. Then, the detected anomalies are correlated with candidate attack paths and verified against a curated TTP (tactics, techniques, and procedures) rule base. Finally, LLMs operate in parallel to generate, normalize, and continuously update TTP rules from threat intelligence and system telemetry, ensuring that the detection process remains adaptive to evolving attacker behaviors. Extensive experiments on a hybrid dataset that combines curated APT samples from five public repositories with a high-fidelity emulated APT scenario corpus show that LLM-HGAN achieves 95% detection accuracy, surpasses state-of-the-art baselines by at least 14%, and reaches 98% accuracy on real-world APT variants, demonstrating strong robustness and generalization to unknown-vulnerability attacks.

The cyber physical systems of smart cities are facing increasingly severe attack situations, and traditional separate protection methods are difficult to effectively respond to. It is urgent to coordinate public security and network security protection. However, the integration of the two faces many challenges and is a very promising research field. The aim of this study is to investigate technical approaches for the synergy between public safety and cyber security. This paper proposes a smart city safety protection model inspired by the human immune mechanism. It was found that through a three line defense architecture similar to the human immune mechanism, and with the help of certain algorithms and functional middleware modules, public security and network security protection components can be dynamically combined to achieve collaboration. This work has verified through experiments a promising path to effectively resist complicated attack threats intertwined with public safety and cyber security factors.