Model3h ago
Claude Mythos Preview Finds Thousands Of Vulnerabilities
Thousands of high-severity vulnerabilities in major operating systems and web browsers surfaced via Anthropic's Claude Mythos Preview. The model identified these flaws without explicit training for security research. This capability forces Frontier Red Team to rethink how developers secure code against AI-driven discovery. Practitioners must now assume zero-day exploits will scale.