OpenAI says it has removed the “warning” messages in its AI-powered chatbot platform, ChatGPT, that indicated when content might violate its terms of service.