The deformation monitoring of super high-rise buildings is crucial to the stability of the building structure and is an indispensable step in ensuring building safety. However, the current building deformation monitoring methods have problems such as poor data collection quality and poor ability to capture complex structures and feature relationships in images, making it difficult to ensure monitoring accuracy. Therefore, this study proposes a deformation monitoring method for super high-rise buildings based on graph neural network. After generating three-dimensional images of super high-rise building structure by three-dimensional laser scanning technology, the images are enhanced and denoised. Afterwards, with the help of graph neural network technology, we deeply studied and analyzed the building images, accurately identified whether there was deformation in the building structure, and determined the amount of deformation. Through experimental verification, it was found that the monitoring values obtained using the method described in this article are highly consistent with the actual values, with a maximum difference of only 1mm. The confidence level of the monitoring results remains above 97%. This method can provide a reference basis for deformation monitoring of super high-rise buildings.