Abstract Low‐cloud feedback continues to dominate uncertainty in climate sensitivity, with no direct observational estimate to date. Here we provide the first such estimate by combining satellite and reanalysis data sets within a cloud‐controlling factor (CCF) framework. The CCF method reproduces observed variability (r=0.93) $(r=0.93)$ and enables reconstruction of low‐cloud cover (LCC) anomalies back to 1979. Contrary to long‐standing expectations, tropical LCC has increased by 0.18% $0.18%$ per decade since 1979, primarily driven by stratocumulus clouds. This corresponds to a strongly negative low‐cloud feedback −0.79Wm−2K−1 $left(-0.79,mathrm{W},{mathrm{m} }^{-mathrm{2} },{mathrm{K} }^{-mathrm{1} }right)$ over the historical record, in contrast with a positive feedback from CMIP6 abrupt‐4×CO2 ${times} {mathrm{CO} }{2}$ projections. Our results provide observational evidence that historical warming has been associated with a negative low‐cloud feedback. If this relationship persists under future warming, it would favor lower climate sensitivity than recent assessments. Counterintuitively, models that best capture observed LCC climatology and trends exhibit the strongest positive abrupt‐4×CO2 ${times} {mathrm{CO} }{2}$ feedback and highest climate sensitivity.