Customers are more uncivil to machines than to other humans. Powered by AI processes, a new phenomenon has emerged: chatbots confronting humans’ uncivil comments. We tested customers’ reactions to chatbots confrontations in three experimental studies. When customers are directly confronted, they find a confrontation from a chatbot less appropriate than from humans because they believe that chatbots do not have moral agency. When they are observers, they find a confrontation from a chatbot as appropriate as from humans. We intend to conduct other experiments to investigate when and why AI can be allies.