Abstract Climate models with identical anthropogenic forcings simulate different historical warming, mainly because aerosol forcing is poorly constrained. The role of uncertainty in regional aerosol forcing, however, has not been systematically studied. Using a large ensemble of CMIP6 models, I demonstrate that anthropogenic aerosol‐forced North Atlantic (NA) sea‐surface temperature variability significantly modulates greenhouse gas‐driven global‐mean surface air temperature (GMST) increases in historical simulations, explaining 6.8 (±0.7)% of the ensemble‐mean GMST change. This effect varies widely across models. Consequently, models with strong NA forcing tend to underestimate anthropogenic global warming, and vice‐versa, independent of climate sensitivity. The regional forcing variation is linked to a coherent hemispheric pattern, with the largest variations occurring near the anthropogenic aerosol source regions. Our results highlight the importance of regional aerosol forcing in understanding historical warming and, potentially, future projections, given the significant uncertainty in model simulations.