Background: Encephalitis is a severe and potentially life-threatening inflammatory disorder of the central nervous system. Without prompt diagnosis and appropriate treatment, it often results in poor clinical outcomes. The study aimed to develop an artificial intelligence-based model that distinguishes autoimmune encephalitis from infectious encephalitis, encompassing a broad spectrum of autoimmune encephalitis phenotypes, serostatuses, and neuroimmunological entities. Methods: We conducted a retrospective analysis of patients diagnosed with autoimmune encephalitis, including paraneoplastic neurological syndromes and/or infectious encephalitis, at Vilnius University Hospital Santaros Klinikos from 2016 to 2024. Supervised machine learning techniques were used to train the models, and Shapley Additive Explanations analysis was applied to improve their interpretability. Results: A total of 233 patients were included in the study. The Random Forest model demonstrated the best performance in differentiating the etiology of encephalitis, achieving an AUROC of 0.966. Further analysis revealed that laboratory, electroencephalography, and clinical data were the most influential predictors, whereas imaging data contributed less to classification accuracy. Conclusions: We developed a machine learning model capable of distinguishing infectious encephalitis from both seropositive and seronegative autoimmune encephalitis. Since autoimmune cases may be misdiagnosed as infectious in the absence of detectable antibodies, our model has the potential to support clinical decision-making and reduce diagnostic uncertainty.