AI Impact HubAnthropic's Project Glasswing finds 23,019 vulnerabilities across 1,000-plus open-source projects
Anthropic published the first wave of results from Project Glasswing, the restricted initiative that gives partners including AWS, Apple, Cisco, Google, JPMorgan Chase, Microsoft, NVIDIA, Broadcom, CrowdStrike, Palo Alto Networks and the Linux Foundation pre-release access to Claude Mythos. Scanning more than 1,000 open-source projects, Mythos identified 23,019 issues, 6,202 of them high or critical severity, including flaws in every major operating system and web browser. Anthropic frames this as evidence frontier models have crossed the threshold where they can outperform all but the most skilled humans at finding and exploiting software vulnerabilities.
Why it matters: AI is now a first-class tool on both sides of the cyber arms race, and the partner list shows the biggest software vendors are already plugged in before public release.
helpnetsecurity.com