The paper focuses on the study of Trojan horse malware. The paper discusses about a brief introduction of Trojan horse malware, its types, how and where the Trojan horse can attack. Along with the ways to attack, the paper will be discussing about different machine learning models trained and available till now and about their accuracy too. The survey of 10 + research papers has been carried out to get updated with the latest advancements in this field and to study different machine learning algorithms to detect Trojan horse malware with as high accuracy as possible. The paper will be discussing about the attack of Trojan horse in different systems like Android, PC, hardware, etc. Considering the survey of recent papers from 2020–2023, it has been observed that the machine learning model can judge the Trojan horse according to its signature and behaviour. The paper will explore the ways to extract features like MD5 hash value of a file or its behaviour to detect Trojan horse. The paper will also discuss about different classifiers like random forest, decision tree, J48, etc. to classify the Trojan horse malware and will be comparing their accuracy rate and choose the best. Signature-based analysis is not well enough for detection, instead behavioural-based analysis gives more accuracy by examining the actions and interactions of malware within a system. At the end, the paper will be discussing about the challenges and limitations associated with Trojan horse malware detection using machine learning.