AI painting is the borrowing of art to technology, and there is a risk that technology holds art hostage and causes the dual alienation of art and technology, which is characterized by AI painting server. In order to avoid the improper application of AI technology in the field of painting, it is necessary to analyze the risk from the perspective of service failure. Aiming at the problems of ignoring uncertainty fuzzy evaluation in traditional FMEA, a Fuzzy-FMEA method introducing fuzzy language evaluation was proposed to identify the risks of AI painting services and determine the priority of rectification. Firstly, the experience experiment was designed, the AI painting service contact points were sorted out, the complaint response of the subjects was recorded during the experience, and the complaint index was calculated, which laid the foundation for the subsequent Fuzzy-FMEA. Secondly, an FMEA team was organized to establish a fuzzy language evaluation set, perform failure analysis on service contacts that affected the service efficiency of agents due to improper application of AI technology, calculate the failure risk value, and determine the priority of measures according to the 5th edition of the FMEA manual. Combined with experimental data, the early warning reports on risks and reorient services are output and the feasibility and effectiveness of the proposed method are verified by specific cases.