On May 24, 2026, Anthropic in San Francisco released the first performance report for Project Glasswing, a collaborative cybersecurity initiative using a specialized Claude AI model. During its first 30 days, the system identified more than 10,000 high- or critical-severity software vulnerabilities across partner infrastructure and applications. The report states that these issues include memory corruption, injection flaws, and unauthorized access vectors, most of which had not been detected by existing enterprise scanning tools. Project Glasswing is built on Claude Mythos Preview and involves major partners such as AWS, Apple, Cisco, CrowdStrike, Google, JPMorganChase, Microsoft, NVIDIA, Palo Alto Networks, and the Linux Foundation.

Prepared by Jonathan Pierce and reviewed by editorial team.