Rivers in high latitude regions are used for fishing, hydroelectric power generation, recreation, transporting supplies to isolated communities and hold cultural significance for many indigenous communities. To model how annual river ice breakup timing is likely to change under future climate change scenarios, we selected daily meteorological variables from 15 CMIP6 Earth system models (ESMs), in our study conducted across 40 locations across interior Alaska. Observed river ice breakup records for the 40 locations were retrieved from the Alaska-Pacific River Forecast Center. A long short-term memory model (LSTM) designed to predict singular annual events, was tuned, trained and evaluated on the historical simulation of each ESM, using the daily climate variables as inputs to predict annual river ice breakup timing. A mean ensemble using 15 LSTMs was created, with a mean absolute percentage error (MAPE) of 7.09% (test data corresponding to 1980 - 2014). The ensemble was applied to five future scenarios: SSP1-1.9, SSP2-4.5, SSP3-7.0, SSP5-8.5, SSP5-3.4-overshoot; to forecast future river ice breakup timing. Using Theil-Sen estimation and the Mann-Kendall test for monotonic trend, we conclude that breakup timing is predicted to occur earlier in the Spring season under all scenarios except SSP1-1.9, with scenarios with higher CO2 concentrations showing earlier breakup timing.