since openAI and Anthropic are public about their anti-distillation mechanism, its either their anti-distillation doesn't work or kimi got here because their own training and RLs. Which one is it?