Artificial intelligence (AI) is transforming urban systems, yet its influence on persistent challenges like carbon emissions and its effects across regional boundaries require urgent clarification for effective climate action. Based on panel data from 284 Chinese cities between 2005 and 2021, this study utilizes the Spatial Durbin Model to quantify the impact of AI on local and neighboring city carbon emissions. Empirical findings reveal that AI significantly suppresses emissions in both areas. Specifically, under the collaborative innovation weight matrix, the estimations demonstrate that AI development reduces local carbon emission intensity by a significant coefficient of 0.0594 and neighboring emission intensity by 0.0183. Furthermore, active participation in collaborative innovation networks substantially amplifies these emission reduction effects. Geographical distance yields a nonlinear spillover pattern, where carbon-increasing externalities dominate within the 100 km threshold but attenuate rapidly at greater distances. Mechanism tests confirm that AI operates by reducing energy consumption intensity, boosting green technology innovation, and promoting service industry agglomeration. The research highlights the critical potential of AI, particularly when efficiently integrated into regional innovation networks, to mitigate challenging administrative boundary pollution and to inform highly strategic sustainable urban development pathways.