Huh, that's an interesting and new perspective. I'd love to know what you mean by safety systems, and what OpenAI can do that Anthropic can't.