Because consumers of information now frequently utilize online social media sites for communication, malicious attacks that target these kinds of networks are also increasing. In this instance, our objective is to create a hybrid customized system that detects spyware using a fuzzy system method and artificial intelligence. When important data is uploaded or maintained on social media sites, malicious attacks have the potential to entirely modify or distort it, harming both individuals and organizations. To perform the fuzzy rules derivation and validate the efficacy of the hybrid approach, the combined approach being subjected to identification of malware tests which have been made available in many public datasets. In binary categorization tests, it was compared against hybrid fuzzy neural network designs and models of artificial neural networks. The consequences of the simulation demonstrate the viability of the fuzzy neural networks methods to identifying malware and its ability to create fuzzy rules that can aid in the creation of customized systems. Next, we improve it via the Text Mining as well as Op code-based learning approach using the IPS-MD5 technique to prevent the spread of malicious programs in social networking sites. As consequence, there will be fewer malevolent attacks.