OpenAI unveils 'Aardvark,' a GPT-5-powered agent for autonomous cybersecurity research

OpenAI launched Aardvark, a GPT-5-powered cybersecurity agent that autonomously discovers, explains, and helps fix security vulnerabilities in code repositories. Currently in private beta with select partners, the tool examines codebases to understand their purpose, scans for vulnerabilities, tests them in sandboxed environments, and generates patches for human review. Originally developed as an internal OpenAI tool, Aardvark addresses the challenge of thousands of new vulnerabilities discovered annually across enterprise software.