The Anthropic model identified 271 security vulnerabilities in the Firefox browser.

Anthropic's latest AI model, Claude Mythos, identified 271 security vulnerabilities in Mozilla's Firefox browser during testing. These vulnerabilities have recently been fixed, highlighting the potential of AI in cybersecurity. Mozilla stated that the model can efficiently analyze large-scale codebases and locate potential vulnerabilities, a process that previously relied heavily on scarce, highly skilled security researchers. In previous tests, another Claude model discovered 22 critical vulnerabilities. Nevertheless, Mozilla believes that the vulnerabilities discovered by AI remain within the capabilities of top human researchers, but its scalability could potentially change the landscape of attack and defense, giving defenders a systemic advantage for the first time. However, industry experts also warn that the misuse of this technology could exacerbate the risk of automated attacks. (Decrypt)