US National Security Agency reportedly using Anthropic's Mythos AI model for hacking

The NSA is apparently using Mythos, Anthropic's unreleased AI model, for offensive hacking purposes, even as the US government fights a legal battle against Anthropic. Mythos has not been publicly released because of its powerful abilities to detect and exploit software vulnerabilities.

Mythos is instead being offered to select institutions to help them build defenses against such exploits. Meanwhile, Cloudflare reports that most web traffic is now AI-led, arriving a year ahead of earlier projections, with a decent minority being malicious bots.

Top open-source reasoning models are less than a year behind the frontier systems like Mythos, though they currently lack its cutting-edge capabilities for vulnerability detection and exploitation.