#AI cybersecurity vulnerabilities

Anthropic's Mythos AI found thousands of unknown software vulnerabilities. But cybersecurity experts say the same capability already exists in older, publicly available models — and defenses are nowhere near keeping up.

Anthropic's tightly restricted Mythos AI—designed to find security flaws—was accessed by Discord sleuths without a single line of exploit code. Meanwhile, North Korean hackers used AI to steal $12M in three months. The security paradox of 2026.