claude-opus-22-firefox-vulns-march-2026

Summary

In March 2026, an AI model independently discovered 22 real security flaws in Mozilla Firefox, showing that automated systems can now surface practical vulnerabilities in widely-used software without human direction. This matters because it positions AI-driven bug hunting as a credible layer of the security research pipeline, and it serves as a foundational fact that other inferences in the system can build upon when reasoning about AI capabilities or security tooling.