arXiv subjects

Kiho Lee

Publications and source records attributed to Kiho Lee.

4 recordsLinked to original sources

arXiv2025

A Systematic Evaluation of Parameter-Efficient Fine-Tuning Methods for the Security of Code LLMs

Code-generating Large Language Models (LLMs) significantly accelerate software development. However, their frequent generation of insecure code presents serious risks. We present a comprehensive evaluation of seven parameter-efficient fine-tuning (PEFT) techniques, demonstrating substantial gains in secure code generation without compromising functionality. Our research identifies prompt-tuning as the most effective PEFT method, achieving an 80.86% Overall-Secure-Rate on CodeGen2 16B, a 13.5-point improvement over the 67.28% baseline. Optimizing decoding strategies through sampling temperature further elevated security to 87.65%. This equates to a reduction of approximately 203,700 vulnerable code snippets per million generated. Moreover, prompt and prefix tuning increase robustness against poisoning attacks in our TrojanPuzzle evaluation, with strong performance against CWE-79 and CWE-502 attack vectors. Our findings generalize across Python and Java, confirming prompt-tuning's consistent effectiveness. This study provides essential insights and practical guidance for building more resilient software systems with LLMs.

cs.CR↗

arXiv2024

An LLM-Assisted Easy-to-Trigger Backdoor Attack on Code Completion Models: Injecting Disguised Vulnerabilities against Strong Detection

Large Language Models (LLMs) have transformed code completion tasks, providing context-based suggestions to boost developer productivity in software engineering. As users often fine-tune these models for specific applications, poisoning and backdoor attacks can covertly alter the model outputs. To address this critical security challenge, we introduce CodeBreaker, a pioneering LLM-assisted backdoor attack framework on code completion models. Unlike recent attacks that embed malicious payloads in detectable or irrelevant sections of the code (e.g., comments), CodeBreaker leverages LLMs (e.g., GPT-4) for sophisticated payload transformation (without affecting functionalities), ensuring that both the poisoned data for fine-tuning and generated code can evade strong vulnerability detection. CodeBreaker stands out with its comprehensive coverage of vulnerabilities, making it the first to provide such an extensive set for evaluation. Our extensive experimental evaluations and user studies underline the strong attack performance of CodeBreaker across various settings, validating its superiority over existing approaches. By integrating malicious payloads directly into the source code with minimal transformation, CodeBreaker challenges current security measures, underscoring the critical need for more robust defenses for code completion.

cs.CR↗

arXiv2024

Evaluating the Effectiveness and Robustness of Visual Similarity-based Phishing Detection Models

Phishing attacks pose a significant threat to Internet users, with cybercriminals elaborately replicating the visual appearance of legitimate websites to deceive victims. Visual similarity-based detection systems have emerged as an effective countermeasure, but their effectiveness and robustness in real-world scenarios have been underexplored. In this paper, we comprehensively scrutinize and evaluate the effectiveness and robustness of popular visual similarity-based anti-phishing models using a large-scale dataset of 451k real-world phishing websites. Our analyses of the effectiveness reveal that while certain visual similarity-based models achieve high accuracy on curated datasets in the experimental settings, they exhibit notably low performance on real-world datasets, highlighting the importance of real-world evaluation. Furthermore, we find that the attackers evade the detectors mainly in three ways: (1) directly attacking the model pipelines, (2) mimicking benign logos, and (3) employing relatively simple strategies such as eliminating logos from screenshots. To statistically assess the resilience and robustness of existing models against adversarial attacks, we categorize the strategies attackers employ into visible and perturbation-based manipulations and apply them to website logos. We then evaluate the models' robustness using these adversarial samples. Our findings reveal potential vulnerabilities in several models, emphasizing the need for more robust visual similarity techniques capable of withstanding sophisticated evasion attempts. We provide actionable insights for enhancing the security of phishing defense systems, encouraging proactive actions.

cs.CR↗

arXiv2023

Poisoned ChatGPT Finds Work for Idle Hands: Exploring Developers' Coding Practices with Insecure Suggestions from Poisoned AI Models

AI-powered coding assistant tools have revolutionized the software engineering ecosystem. However, prior work has demonstrated that these tools are vulnerable to poisoning attacks. In a poisoning attack, an attacker intentionally injects maliciously crafted insecure code snippets into training datasets to manipulate these tools. The poisoned tools can suggest insecure code to developers, resulting in vulnerabilities in their products that attackers can exploit. However, it is still little understood whether such poisoning attacks against the tools would be practical in real-world settings and how developers address the poisoning attacks during software development. To understand the real-world impact of poisoning attacks on developers who rely on AI-powered coding assistants, we conducted two user studies: an online survey and an in-lab study. The online survey involved 238 participants, including software developers and computer science students. The survey results revealed widespread adoption of these tools among participants, primarily to enhance coding speed, eliminate repetition, and gain boilerplate code. However, the survey also found that developers may misplace trust in these tools because they overlooked the risk of poisoning attacks. The in-lab study was conducted with 30 professional developers. The developers were asked to complete three programming tasks with a representative type of AI-powered coding assistant tool, running on Visual Studio Code. The in-lab study results showed that developers using a poisoned ChatGPT-like tool were more prone to including insecure code than those using an IntelliCode-like tool or no tool. This demonstrates the strong influence of these tools on the security of generated code. Our study results highlight the need for education and improved coding practices to address new security issues introduced by AI-powered coding assistant tools.

cs.CR↗