Anthropic is finding bugs faster than Microsoft can fix them

Anthropic has given select software makers, including Microsoft, access to a new AI model called Mythos that uncovers security vulnerabilities in code at an unprecedented pace. Per Ars Technica, the capability prompted an urgent gathering in mid-May, when Microsoft engineers convened in Redmond to confront a critical challenge: Mythos was finding flaws in their code faster than the company could patch them. The race has a name — Project Glasswing — and a genuine threat model: the goal is to seal vulnerabilities before adversarial nations like China use similar AI tools to exploit them for espionage or sabotage.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton