Caleb Gross Firmware Function Ranker
A research-based approach using Large Language Models to rank decompiled firmware functions by their likelihood of containing security vulnerabilities.
0 projects
About Caleb Gross Firmware Function Ranker
Developed by Caleb Gross and presented at Black Hat USA 2026, this methodology addresses the challenge of auditing firmware containing thousands of decompiled functions without symbol information. The approach utilizes a Large Language Model (LLM) to prioritize functions based on their potential to hide bugs. According to the research, this ranking method more than doubled the precision compared to simply asking the model for direct identification. The tool is designed to assist security researchers in navigating large, unlabeled codebases by focusing manual review efforts on the most high-risk areas of the firmware.
Summarised from the official page linked above.