Abstract: A social network opinion model is proposed to address a type of public opinion security issue in social networks, which includes suspects, defenders, and ordinary individuals. Suspects and defenders are modeled as network nodes with external inputs. Suspects spread malicious opinions to other individuals, while defenders spread positive opinion to other individuals. The safety of public opinion is judged by comparing the distance between the expected opinions of the group and the malicious and positive opinions. Based on this, the parameter conditions that should be met to maintain the safety of public opinion are determined. Furthermore, the sufficient conditions for the convergence of expected opinions of social groups are analyzed, and a calculation method for designing low-cost, effective positive opinion is designed. Based on theoretical analysis and experimental verification, the following conclusions are obtained. If there is no defender in social networks, suspects can cause public opinion crises by spreading malicious opinion. When a public opinion crisis erupts, all individuals in the social network hold malicious opinion. If there are defenders in social networks, they can prevent public opinion crisis by spreading positive opinion. In this case, suspects in the social network can be detected according to abnormal update patterns of their opinions, further ensuring the safety of public opinion. In addition, it can be found that the number of suspects and defenders also affects the security level of public opinion. The more suspects, the greater the likelihood of triggering a public opinion crisis. The more defenders, the more likely to prevent the outbreak of public opinion crisis.