Anthropic’s AI Finds Bugs Faster Than Microsoft Can Patch Them

In mid-May, Microsoft engineers gathered at its Redmond headquarters to address Project Glasswing, an effort to fix security weaknesses uncovered by Anthropic’s AI model, Mythos. The model had been shared with select organizations to identify vulnerabilities in widely used software before malicious actors could exploit them. During the meeting, an engineer questioned whether Mythos lived up to Anthropic’s claims.