openai: "Our models are more dangerous"

anthropic: "No, our models are more dangerous"