OpenAI launched Aardvark, a GPT-5-powered autonomous security tool that discovers and addresses software vulnerabilities at scale. Unlike traditional fuzzing or composition analysis, Aardvark uses LLM reasoning to analyze code repositories, create threat models, monitor commits, and validate vulnerabilities in sandboxed environments. In benchmark tests, it identified 92% of known vulnerabilities and discovered multiple CVEs in open-source projects. The tool integrates with GitHub and Codex, automatically generating patch proposals for human review. Currently in beta with select partners.
1 Comment
Sort: