In this talk, the use of generative neural networks (GNNs) for dependence modeling in place of parametric dependence models is explored. Particularly, a type of GNN known as the generative moment matching network (GMMN) is investigated in the contexts of two dependence modeling tasks. First, GMMNs are introduced for generating quasi-random samples from multivariate distributions with any underlying dependence structure. These GMMN quasi-random samples are then used to obtain low-variance estimates of quantitative risk measures of interest. Second, a GMMN--GARCH approach is proposed for modeling dependent multivariate time series, where ARMA--GARCH models are utilized to capture the temporal dependence within each univariate time series and GMMNs are used to model the underlying cross-sectional dependence. The primary objective of our proposed approach is to produce probabilistic forecasts which in turn can be used to forecast various risk measures.