OpenAI releases Aardvark, a bug-finding AI agent that automatically reads code, finds vulnerabilities, and writes fixes.

OpenAI has released Aardvark, a GPT-5 powered agent that automatically discovers and fixes security vulnerabilities in large-scale codebases. Aardvark can identify 92% of known and manually injected vulnerabilities and can also locate issues under complex conditions. By monitoring code commits, analyzing and verifying vulnerabilities, and generating remediation suggestions, it is deeply integrated with OpenAI Codex to generate patches for vulnerabilities.