AI model Mythos uncovers hundreds of serious Firefox bugs, boosting software security efforts.

Anthropic's Mythos AI model has identified hundreds of high-severity bugs in Mozilla Firefox, including some dormant for over a decade, marking a major advance in AI-driven software security. Mozilla's researchers report a dramatic increase in bug fixes, from 31 to 423 in one year, thanks to Mythos' ability to detect complex vulnerabilities like sandbox exploits more effectively than human researchers. While AI assists in coding patches, human engineers still review and finalize fixes. The development signals a potential shift in cybersecurity, with AI tools enhancing defenders' capabilities, though the full impact remains uncertain as some bugs await patching.