Anthropic reveals large-scale distillation attacks targeting Claude.

Millions of structured prompts were used to extract reasoning, coding, and agentic capabilities.

AI model extraction is now a serious cybersecurity and national security threat.