Only 200 partner organizations can access Mythos, a new tool from Anthropic. The company blocked a general release because the AI identifies software vulnerabilities with dangerous precision. This restriction prevents bad actors from automating data theft or infrastructure attacks. Security practitioners must now rely on limited partnerships to utilize the tool.