Project Glasswing by Anthropic didn't just find the bugs. It also found the real vulnerability in cybersecurity.

Anthropic has released details from Project Glasswing, an initiative using its Claude Mythos AI model to identify cybersecurity vulnerabilities. According to the announcement, Claude Mythos found over 10,000 critical vulnerabilities in a single month. However, of the 6,200 vulnerabilities identified in open-source software, only 75 have been patched so far.

The project highlights the growing capability of AI models to automate vulnerability discovery at scale. Anthropic and external cybersecurity researchers have noted that the bottleneck has shifted from finding bugs to fixing them, as the volume of discovered vulnerabilities outpaces the capacity of open-source maintainers to issue patches. The 75 patches represent a small fraction of the total.

The findings raise questions about the security of widely used open-source software, which underpins many commercial and government systems.