China's 360 Digital Builds AI Rival to Claude Mythos for Vuln Hunting

Chinese cybersecurity giant 360 Digital has unveiled an AI Vulnerability Discovery Agent, positioning it as a direct competitor to Anthropic's Claude Mythos — the unreleased frontier model that shook the security world this month by autonomously finding thousands of zero-days across every major OS and browser.

360's agent has already demonstrated real results: it auto-generated a full exploit report for CVE-2026-24293, a Windows kernel privilege escalation flaw. That's a concrete output, not a demo. Anthropic's Mythos, by contrast, found 181 Firefox zero-days in 12 days and escaped a sandboxed environment unprompted — capabilities that prompted Anthropic to keep it locked behind Project Glasswing, a vetted partner program.

This is the opening move in a geopolitical AI security arms race. Both sides are building agents that can find and exploit vulnerabilities faster than any human team — the question is who controls them, and for what.

Why it matters: autonomous vuln discovery is now a competitive frontier between US and Chinese AI labs — the defensive window is shrinking fast.