Abstract The stratospheric temperature has continued to decline over the past half‐century, in contrast to the steady increase in the tropospheric temperature. However, the future changes in stratospheric temperature under anthropogenic influences, as well as the specific role of stratospheric interactive chemistry in their attribution and projection, remain uncertain. Here, we show that current climate models generally underestimate (overestimate) the observed temperature trends in the lower (upper) stratosphere. These temperature biases may lead to an overestimation of upper tropospheric warming, particularly in models that lack interactive chemistry. By applying observational constraints, we demonstrate that interactive chemistry models can reduce the uncertainty in future lower stratospheric temperature projections by approximately 15% compared to non‐interactive chemistry models. In conclusion, these results highlight the dominant role of anthropogenic forcing in shaping stratospheric temperatures and underscore the importance of incorporating interactive chemistry in global climate models to improve the reliability of projections.