Mozilla has patched 271 security vulnerabilities in Firefox 150 using early access to Anthropic's Mythos Preview AI model. Firefox CTO Bobby Holley described the tool as every bit as capable as the world's best security researchers. The foundation says the AI helps defenders gain an edge in cybersecurity.
Mozilla announced in a blog post that Anthropic's Mythos Preview model identified 271 vulnerabilities in the source code of Firefox 150, which was released this week. The tool analyzed the unreleased code, enabling the team to fix issues before launch. Firefox CTO Bobby Holley wrote that the vulnerabilities could have been found through automated fuzzing or by elite researchers, but Mythos did so efficiently, saving months of human effort in many cases. He noted that Anthropic's previous Opus 4.6 model found only 22 bugs in Firefox 148 last month, highlighting Mythos's improvement. Holley stated, “Computers were completely incapable of doing this a few months ago, and now they excel at it. We have many years of experience picking apart the work of the world’s best security researchers, and Mythos Preview is every bit as capable.” Mozilla emphasized that no vulnerability category or complexity evades the model if humans can find it. In a Wired interview, Holley predicted that AI-aided analysis will become standard for all software, as bugs are now easily discoverable. He added that Mozilla has a head start and has “rounded the curve,” tilting the balance toward defenders, especially for open-source projects. The Firefox team described adjusting to the volume of bugs as a necessary challenge, given that such capabilities will soon reach attackers. Firefox users can disable related AI features if desired.
