AI sandboxing just proved what Kubernetes security has been missing

Anthropic's Mythos model recently did something the security industry would prefer not to think about too hard: it autonomously chained zero-day vulnerabilities across every major OS and browser, including a 27-year-old bug that survived decades of human review. No specialized training. No human gui