Anthropic's Mythos AI Finds Thousands of Zero-Day Bugs
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Anthropic announced Project Glasswing, a cybersecurity initiative using its Claude Mythos Preview model to find and fix software vulnerabilities in major platforms.
- Claude Mythos Preview identified thousands of zero-day vulnerabilities in every major operating system and web browser, with some flaws reportedly unpatched for decades, per Tom's Hardware.
- Anthropic deemed the Mythos model too dangerous to release publicly, restricting access to participating security researchers.
- Project Glasswing brings together an unusual cross-industry coalition including Apple, AWS, Broadcom, Cisco, NVIDIA, and Microsoft, with Apple and Amazon receiving access to test the more powerful Mythos model.
- Anthropic is adding compute partners including Google and Broadcom to support the initiative's scale, per CIO Dive.
Why it matters: The discovery of vulnerabilities unpatched for decades means major OSes and browsers have harbored known-severity flaws in the wild, and Anthropic's refusal to release the model publicly reflects how capable it is at finding attack paths. The cross-industry coalition — with Apple, Amazon, Google, Microsoft, and Cisco all participating — signals AI-powered vulnerability hunting has become urgent enough to unite normally rival tech giants around a shared defensive effort.
