AI-driven cybersecurity tools transform defense strategies

Anthropic has introduced Methus, an AI model that can automatically discover and fix software vulnerabilities in critical systems such as Linux, OpenBSD and FFmpeg. Rather than commercialising the technology for attackers, Anthropic offers it to defenders through Project Glass Wing, allowing major providers including AWS, Apple, Google and Microsoft to patch faster and extending protection to smaller organisations.

Security experts warn that relying on a single, consolidated AI platform can create a systemic single point of failure. They advocate a hybrid AI architecture that centralises AI‑driven analytics and detection while keeping essential functions—such as identity and access management and backup/recovery infrastructure—isolated. This approach aims to combine the speed of AI remediation with resilience against cascading failures.