Airborne laser scanning data (ALS) are used mainly for creation of precise digital elevation models. However, it appears that the informative potential stored in ALS data can be also used for updating spatial databases, including the Database of Topographic Objects (BDOT10k). Typically, geometric representations of buildings in the BDOT10k are equal to their entities in the Land and Property Register (EGiB). In this study ALS is considered as supporting data source. The thresholding method of original ALS data with the use of the alpha shape algorithm, proposed in this paper, allows for extraction of points that represent horizontal cross section of building walls, leading to creation of vector, geometric models of buildings that can be then used for updating the BDOT10k. This method gives also the possibility of an easy verification of up-to-dateness of both the BDOT10k and the district EGiB databases within geometric information about buildings. For verification of the proposed methodology there have been used the classified ALS data acquired with a density of 4 points/m². The accuracy assessment of the identified building outlines has been carried out by their comparison to the corresponding EGiB objects. The RMSE values for 78 buildings are from a few to tens of centimeters and the average value is about 0,5 m. At the same time for several objects there have been revealed huge geometric discrepancies. Further analyses have shown that these discrepancies could be resulted from incorrect representations of buildings in the EGiB database.