Anthropic claims Chinese AI firms distilled Claude to train rival AI models, raising concerns about model extraction, security risks, and AI distillation abuse.