Kun Li

Kun Li, Fan Zhang, Wei Guo

Deep learning technology has made great achievements in the field of image. In order to defend against malware attacks, researchers have proposed many Windows malware detection models based on deep learning. However, deep learning models are vulnerable to adversarial example attacks. Malware can generate adversarial malware with the same malicious function to attack the malware detection model and evade detection of the model. Currently, many adversarial defense studies have been proposed, but existing adversarial defense studies are based on image sample and cannot be directly applied to malware sample. Therefore, this paper proposes an adversarial malware defense method based on adversarial training. This method uses preprocessing to defend simple adversarial examples to reduce the difficulty of adversarial training. Moreover, this method improves the adversarial defense capability of the model through adversarial training. We experimented with three attack methods in two sets of datasets, and the results show that the method in this paper can improve the adversarial defense capability of the model without reducing the accuracy of the model.

Jingyan Zhou, Kun Li, Junan Li et al.

Existing efforts in safeguarding LLMs are limited in actively exposing the vulnerabilities of the target LLM and readily adapting to newly emerging safety risks. To address this, we present Purple-teaming LLMs with Adversarial Defender training (PAD), a pipeline designed to safeguard LLMs by novelly incorporating the red-teaming (attack) and blue-teaming (safety training) techniques. In PAD, we automatically collect conversational data that cover the vulnerabilities of an LLM around specific safety risks in a self-play manner, where the attacker aims to elicit unsafe responses and the defender generates safe responses to these attacks. We then update both modules in a generative adversarial network style by training the attacker to elicit more unsafe responses and updating the defender to identify them and explain the unsafe reason. Experimental results demonstrate that PAD significantly outperforms existing baselines in both finding effective attacks and establishing a robust safe guardrail. Furthermore, our findings indicate that PAD excels in striking a balance between safety and overall model quality. We also reveal key challenges in safeguarding LLMs, including defending multi-turn attacks and the need for more delicate strategies to identify specific risks.

Kun Li, Fan Zhang, Wei Guo

Malware detection models based on deep learning have been widely used, but recent research shows that deep learning models are vulnerable to adversarial attacks. Adversarial attacks are to deceive the deep learning model by generating adversarial samples. When adversarial attacks are performed on the malware detection model, the attacker will generate adversarial malware with the same malicious functions as the malware, and make the detection model classify it as benign software. Studying adversarial malware generation can help model designers improve the robustness of malware detection models. At present, in the work on adversarial malware generation for byte-to-image malware detection models, there are mainly problems such as large amount of injection perturbation and low generation efficiency. Therefore, this paper proposes FGAM (Fast Generate Adversarial Malware), a method for fast generating adversarial malware, which iterates perturbed bytes according to the gradient sign to enhance adversarial capability of the perturbed bytes until the adversarial malware is successfully generated. It is experimentally verified that the success rate of the adversarial malware deception model generated by FGAM is increased by about 84\% compared with existing methods.