All screening examinations were randomized 1:1 to either the AI-supported screening or the standard double reading group. There were 53,052 assigned to the AI-supported screening group and 52,882 assigned to the double reading group. Sensitivity in the AI-supported screening was higher compared with standard screening when it came to invasive cancer (78.3% vs 70.9%). AI-supported screening resulted in 12% fewer interval cancers and 16% fewer invasive interval cancers compared with standard screening. The way AI influences mammography screening over time would change, which could also have affected results.