New types of radar signals, which can be complex and vary rapidly during short time periods, have drawn attention to the use of AI/machine learning for electronic support measures (ESM). This literature study aims to investigate the current research status of the area. Around 50 publications, all from 2016 or later, have been reviewed and categorised according to which ESM function they use AI for. The publications are summarised and factors such as which AI method is used, how each dataset is described, the number of available signals for training/testing of the AI algorithm, and if the study is based on real or simulated data, are presented. However, much ESM research is classified and not publicly available. This, combined with the risk of bias regarding which publications could be found, due to keywords, search platforms, et cetera, makes it impossible to produce a comprehensive, representative view of the current research situation. All results presented here must therefore be interpreted with this caveat in mind. The research within this area varies greatly, and investigates how AI can be used for many types of problems, with different AI techniques, and with the problem statement viewed from different angles. The data sets used for training are in most cases simulated and contain only a few emitters, which make the results not directly transferable to real scenarios. This may indicate that the research still is in its early stages and it is not yet determined how and in which situations AI best can be applied. In general the researchers seem to agree that AI methods have a good potential to perform well for ESM-related tasks, and most probably the interest in AI methods will continue to increase in this field of research. The degree of accuracy, choice of AI algorithm, and so on, is, however, specific for each particular task and it is difficult to draw general conclusions.