Anthropic’s big announcement earlier this month about using AI to improve cybersecurity was met with a lot of skepticism. However, Mozilla has shared some details that support the use of the company’s special Cloud Mythos preview model as a way to protect critical services. Using Mythos helped the Mozilla team find and fix 271 vulnerabilities in the latest release of the Firefox browser. “We have not yet found any category or complexity of vulnerabilities that humans can detect that this model cannot detect,” the foundation said.
Mozilla’s blog post seems like a positive sign for Anthropic’s Project Glasswing. Obviously an AI company will want to present itself in the best possible light when presenting its initiative, but there is something encouraging about hearing the benefits from a third party. Mozilla also noted that in its time with Cloud Mythos, the AI was not able to find any bugs that a human could not find, despite considerable time and resources, which indicates that AI is currently no more capable of cracking cybersecurity defenses than a person.
That an organization is successfully using AI for good is certainly a refreshing change of pace for tech news. And for those Firefox users who aren’t personally interested in implementing any generative AI into their browsing, Mozilla has offered the option to turn it off for the past several months.
<a href