Kumar Shashwat, Francis Hahn, Xinming Ou, Dmitry Goldgof, Jay Ligatti, Larrence Hall (University of South Florida), S. Raj Rajagoppalan (Resideo), Armin Ziaie Tabari (CipherArmor)

Large language models (LLM) are perceived to offer promising  potentials for automating security tasks, such as those found in security operation centers (SOCs). As a first step towards evaluating this perceived potential, we investigate the use of LLMs in software pentesting, where the main task is to automatically identify software security vulnerabilities in source code. We hypothesize that an LLM-based AI agent can be improved over time for a specific security task as human operators interact with it. Such improvement can be made, as a first step, by engineering prompts fed to the LLM based on the responses produced, to include relevant contexts and structures so that the model provides more accurate results. Such engineering efforts become sustainable if the prompts that are engineered to produce better results on current tasks, also produce better results on future unknown tasks. To examine this hypothesis, we utilize the OWASP Benchmark Project 1.2 which contains 2,740 hand-crafted source code test cases containing various types of vulnerabilities. We divide the test cases into training and testing data, where we engineer the prompts based on the training data (only), and evaluate the final system on the testing data. We compare the AI agent’s performance on the testing data against the performance of the agent without the prompt engineering. We also compare the AI agent’s results against those from SonarQube, a widely used static code analyzer for security testing. We built and tested multiple versions of the AI agent using different off-the-shelf LLMs – Google’s Gemini-pro, as well as OpenAI’s GPT-3.5-Turbo and GPT-4-Turbo (with both chat completion and assistant APIs). The results show that using LLMs is a viable approach to build an AI agent for software pentesting that can improve through repeated use and prompt engineering.

View More Papers

FirmDiff: Improving the Configuration of Linux Kernels Geared Towards...

Ioannis Angelakopoulos (Boston University), Gianluca Stringhini (Boston University), Manuel Egele (Boston University)

Read More

Beyond the Bytes: Understanding the Limitations of Intrinsic Binary...

Peter Lafosse (Owner and Co-Founder of Vector 35 Inc.)

Read More

MOCK: Optimizing Kernel Fuzzing Mutation with Context-aware Dependency

Jiacheng Xu (Zhejiang University), Xuhong Zhang (Zhejiang University), Shouling Ji (Zhejiang University), Yuan Tian (UCLA), Binbin Zhao (Georgia Institute of Technology), Qinying Wang (Zhejiang University), Peng Cheng (Zhejiang University), Jiming Chen (Zhejiang University)

Read More

AVMON: Securing Autonomous Vehicles by Learning Control Invariants and...

Ahmed Abdo, Sakib Md Bin Malek, Xuanpeng Zhao, Nael Abu-Ghazaleh (University of California, Riverside)

Read More