OpenAI is testing Aardvark, an innovative security tool powered by GPT-5, designed to enhance software security by scanning code for vulnerabilities. Functioning like a security analyst, Aardvark meticulously reviews code repositories, flags potential risks, tests these vulnerabilities in a sandbox environment, and proposes remedial measures. In internal evaluations, Aardvark successfully identified 92% of known vulnerabilities, including those deliberately inserted into the code. The tool has also been applied to open-source projects, revealing multiple issues that subsequently received CVE (Common Vulnerabilities and Exposures) designations. Currently in closed beta, Aardvark is available for use in some internal systems and among selected partners. Developers interested in leveraging this cutting-edge security solution can apply for access. For comparison, Anthropic offers a similar open-source tool for its Claude model, showcasing the growing importance of AI-driven vulnerability detection in software development and cybersecurity.
Source link
Share
Read more