Arctic sea ice plays a pivotal role in shaping the climate system at high latitudes, acting as both an indicator and driver of climate change processes in this sensitive region. Its seasonal variability and long-term decline have far-reaching implications for global climate dynamics, regional ecosystems, and human activities. While climate models indicate clear evidence of human-induced sea ice decline, quantification of the relative contributions of forcing factors in relation to climate-system internal processes remains uncertain. Here, we tackle this uncertainty by employing a combination of statistical analyses on observational data, highlighting the distinct fingerprints of increased atmospheric CO<jats:sub>2</jats:sub> concentration as external forcing, the Atlantic Multidecadal Oscillation (AMO) as well as the North Atlantic Oscillation (NAO), as modes of internal variability, on global sea surface temperature (SST) and Arctic sea ice concentration (SIC) since 1950. Our analyses reveal that rising atmospheric CO2 concentrations are by far the dominant causal factor for SIC variability, while AMO and NAO also play a significant role in either exacerbating or mitigating sea ice loss. Since mid-1980s, the positive trend of the AMO has amplified the declining trend in Arctic sea ice, with its effects being roughly half as large as the effect of rising CO2 concentrations. Linear regression analyses shed light on the physical processes linking the drivers of Arctic sea ice decline both during phases of sea-ice accumulation and melting. Causal links between increasing atmospheric CO2 concentrations, the AMO, the NAO, on the one hand, and observed global SST—Arctic SIC patterns on the other are also established. Observation-based coupled SST-SIC interactions underline the past evolution of Arctic sea ice and emphasize the important roles of these drivers in shaping its current and future evolution.