Top AI Product

Anthropic shared a sweeping update on Project Glasswing — its AI-assisted security testing initiative powered by Claude Mythos. The headline number: more than 10,000 high- or critical-severity vulnerabilities uncovered across widely used software, with several partner organizations reporting bug-discovery rate gains of more than 10x after integrating AI into their testing workflows.

## The coalition

Glasswing launched in April with an unusually broad partner roster: AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks. That’s most of the companies whose software underpins the internet — collaborating on AI-driven vulnerability discovery rather than competing.

## The double edge

The same capability that finds 10,000 bugs for defenders could find them for attackers. Anthropic restricts Claude Mythos to vetted security researchers precisely because an unrestricted model this good at finding exploits is a weapon. Glasswing is the controlled-disclosure version: find the bugs first, patch them across the coalition before they leak.

## Why it matters

This is the clearest real-world evidence yet that frontier models shift the offense-defense balance in security. A 10x bug-discovery gain isn’t incremental — it’s a phase change. The open question is whether coordinated defender programs like Glasswing stay ahead of attackers running the same class of models without the guardrails.