Anthropic's Mythos Preview Uncovers Thousands of OS and Browser Flaws
Anthropic's new general-purpose model found high-severity vulnerabilities across every major OS and web browser.
Anthropic just dropped Mythos Preview, a general-purpose AI model with a serious knack for breaking things. The model discovered thousands of high-severity vulnerabilities spanning every major operating system and web browser.
That's not a typo. Every major OS. Every major browser. Thousands of bugs.
Alongside the model reveal, Anthropic announced Project Glasswing — a collaborative initiative pulling together heavy hitters including Amazon Web Services, Apple, and Broadcom. The project appears designed to coordinate the kind of large-scale security work that Mythos Preview's capabilities make possible.
The vulnerability-hunting prowess positions Mythos Preview as more than just another general-purpose model. It's a signal that AI-driven security research is scaling fast — and finding holes that human researchers either missed or haven't gotten to yet.