Earthquake-induced landslides are considered the third-largest contributor of societal damages caused by an earthquake. Newmark sliding displacement method-based regional seismic landslide hazard assessment is widely used by researchers for identifying vulnerable slopes for a future seismic event. For this purpose, several researchers have proposed regression-based Newmark slope displacement prediction equations based on various ground motion intensity measures and critical acceleration as the slope representative parameter. However, the standard deviation values of these models are significantly high. In this present work, first of its kind, new artificial neural network-based data-driven prediction models for Newmark’s sliding displacement are developed. Different combinations of ground motion intensity parameters (PGA, PGV, Ia, and Tm) and the slope’s critical acceleration value are employed to predict slope displacement. A total of nineteen prediction models (five scalars and fourteen vectors) have been developed using a dataset containing 13,707 slope displacement data points. The ‘efficiency’ and ‘sufficiency’ study of present models reveals that these models exhibit better performance than existing prediction models. A comparative study with existing models shows that the present models are consistent in terms of displacement patterns. The application of the developed prediction model is demonstrated by performing seismic landslide hazard assessment for slopes in Shimla City, India.