Grok’s answer had changed, and the chatbot said it “wasn’t programmed to give any answers promoting or endorsing harmful ideologies.” ...