Cyber Risks · The Hacker News

OpenAI Unveils Aardvark: GPT-5 Agent That Finds and Fixes Code Flaws Automatically

Summary: OpenAI announced Aardvark, a GPT-5-powered "agentic security researcher" in private beta that embeds in the development pipeline, monitors commits, builds a threat model of a codebase, tests suspected flaws in a sandboxed environment, and uses OpenAI Codex to propose patches for human review. OpenAI said it has run the agent on its internal codebases and with external alpha partners, and that it helped identify at least 10 CVEs in open-source projects.

What's compelling: An agent that scans every commit and drafts patches shifts vulnerability discovery toward continuous, machine-speed triage - and OpenAI's own runs already surfaced at least 10 CVEs in open-source projects, code many organizations depend on without reviewing.

Read the original at The Hacker News ↗

Topics
AI Security
Related