Mozilla Used Anthropic’s Mythos to Find and Fix 271 Bugs in Firefox

Mozilla announced that its Firefox 150 browser release includes fixes for 271 vulnerabilities identified using early access to Anthropic's Mythos Preview AI model. This collaboration gave Mozilla a head start in a critical security race, as both Anthropic and OpenAI have recently unveiled AI models with advanced cybersecurity capabilities. Firefox CTO Bobby Holley stated these tools have "changed things dramatically" by enabling automated techniques to cover "the full space of vulnerability-inducing bugs," a feat previously impossible. The move is part of a defensive strategy, acknowledging that these powerful AI bug-hunting capabilities will soon be in attackers' hands.

Holley describes this as a "transitory moment" requiring intense focus, where all software must undergo a security "bootcamp" to find latent bugs now discoverable by AI. He reports that large companies are already pulling thousands of engineers for six-month security sprints. The impact is especially acute for open-source projects and "abandonware," which are widely used but often maintained by small teams or volunteers lacking resources. Mozilla's experience demonstrates both the profound potential of AI for defenders and the urgent, industry-wide challenge it creates, emphasizing the need for coordinated effort and significant resource allocation to secure software in this new era.