OpenAI's new AI tool can find and fix security bugs
OpenAI just rolled out Aardvark, a new AI tool powered by GPT-5 that helps spot and fix security flaws in code.
Originally built for internal use, it's now in private beta with select partners.
The move comes as more software vulnerabilities are being discovered every year, making automated help a big deal.
Aardvark scans code, flags vulnerabilities, and suggests fixes
Aardvark scans code repositories, reviews changes (commits), and flags possible vulnerabilities for humans to check out.
It even tests these issues safely in sandbox environments and uses Codex to suggest fixes you can review before applying.
Tool has already fixed real issues in open-source projects
Unlike old-school tools that rely on manual checks or static analysis, Aardvark combines AI-powered bug detection with automatic patch suggestions.
It's already found real issues in major open-source projects—OpenAI's security team has double-checked and reported them—so this isn't just hype.