AbstractStrong regional differences exist in how hot temperature extremes increase under global warming. Using an ensemble of coupled climate models, we examine the regional warming rates of hot extremes relative to annual average warming rates in the same regions. We identify hot spots of accelerated warming of model‐simulated hot extremes in Europe, North America, South America, and Southeast China. These hot spots indicate where the warm tail of a distribution of temperatures increases faster than the average and are robust across most Coupled Model Intercomparison Project Phase 5 models. Exploring the conditions on the specific day when the hot extreme occurs demonstrates that the hot spots are explained by changes in the surface energy fluxes consistent with drying soils. However, the model‐simulated accelerated warming of hot extremes appears inconsistent with observations, except over Europe. The simulated acceleration of hot extremes may therefore be unreliable, a result that necessitates a reevaluation of how climate models resolve the relevant terrestrial processes.