In the last few years, AI techniques have been applied for vulnerability detection in code. We are developing source code datasets and metrics to accelerate the development of AI-based bug-finding.
The AI Bug Finder is a modular and expandable test bed for evaluating AI-based methods for finding bugs in source code.
The development of the AI Bug Finder is supported by the Department of Homeland Security Science and Technology CAP (press release).
P. Dessauw, A. Delaitre, H. M. Carvalho and V. Okun, "A Modular and Expandable Testbed for Evaluating ML-based Bug Finders," 2022 IEEE 29th Annual Software Technology Conference (STC), Gaithersburg, MD, USA, 2022, pp. 57-58, doi: 10.1109/STC55697.2022.00017.
AI Bug Finder is available for download: GitHub - usnistgov/ai-bugfinder-testbed: A static analyzer powered by AI.
Description of the processing steps: Introduction — AI Bugfinder documentation (nist.gov).