Claude Mythos AI Model Identifies 271 Firefox Vulnerabilities

ME News reports that on April 22 (UTC+8), Mozilla, the developer of the Firefox browser, published a blog post stating that an early version of Anthropic’s Claude Mythos AI model helped identify 271 vulnerabilities within Firefox during internal testing, all of which have been patched this week. The AI model has recently drawn significant attention for its performance in cybersecurity. Mozilla wrote: “As these capabilities become more widely available to defenders, many other teams are now experiencing the same sense of awe we felt when we first saw the test results. For a hardened target, even a single such vulnerability in 2025 would trigger a red alert—so many occurring simultaneously raises the question of whether it’s even possible to keep up.” Previously, Mozilla tested another Anthropic model that identified 22 security-sensitive vulnerabilities in an earlier Firefox release. Mozilla also noted that none of the discovered vulnerabilities exceeded the capabilities of elite human researchers. Mythos, launched in March this year, is Anthropic’s most advanced model for reasoning, coding, and cybersecurity tasks, representing a new tier beyond its earlier Opus series. Pre-release testing showed it could identify thousands of previously unknown vulnerabilities across major operating systems and web browsers. Anthropic restricts access to the system through a limited program called Project Glasswing, granting access only to select tech companies such as Amazon, Apple, and Microsoft. On Monday, it was reported that despite the Trump administration’s directive for federal agencies to cease using Anthropic technology, the U.S. National Security Agency continues to run Claude Mythos Preview on its classified networks. (Source: BlockBeats)