AI Now Finding Real Security Flaws After Years of Failures

After flooding developers with fake bug reports for years, AI models finally got good at finding real security vulnerabilities in early 2026.

The breakthrough could help protect the software running everything from your phone to medical devices.