Update README.md

README.md

@@ -29,4 +29,21 @@ When using this dataset (directly or through GeoBench-2), please cite the origin
 
 > Zhu, Xiao Xiang; Hu, Jingliang; Qiu, Chunping; Shi, Yilei; Kang, Jian; Mou, Lichao; Bagheri, Hossein; Häberle, Matthias; Hua, Yuansheng; Huang, Rong; Hughes, Lloyd; Li, Hao; Sun, Yao; Zhang, Guicheng; Han, Shiyao; Schmitt, Michael; Wang, Yuanyuan. “So2Sat LCZ42: A Benchmark Data Set for the Classification of Global Local Climate Zones.” *IEEE Geoscience and Remote Sensing Magazine*, Vol. 8, No. 3, 2020. https://doi.org/10.1109/MGRS.2020.2964708  
 >
-> Dataset page: https://mediatum.ub.tum.de/1483140  
+> Dataset page: https://mediatum.ub.tum.de/1483140
+
+
+---
+
+## License Disclaimer & Takedown Policy
+
+GEO-Bench-2 redistributes transformed versions of publicly available datasets for research and benchmarking purposes. Each dataset included here is redistributed under the same license terms designated by its original licensor. 
+
+No New Rights Created: ServiceNow Research, IBM Research, TUM, and AI Alliance do not claim ownership over these datasets and do not grant any new rights beyond those already specified by the original licensors. 
+
+Attribution: All credit remains with the original dataset creators and providers. Each dataset is accompanied by license and attribution information pointing to the original source. 
+
+Disclaimer of Responsibility: The consortium partners act only as redistributors. We do not independently verify the accuracy of the license terms supplied by original dataset providers and disclaim liability for any errors, omissions, or misattributions in those designations. 
+
+Takedown Procedure: If you are a rights holder and believe that a dataset has been misattributed, incorrectly licensed, or should not be redistributed in this form, please contact Paolo Fraccaro ([email protected]) and Alexandre Lacoste ([email protected]). We will promptly review and, if necessary, remove or modify access to the dataset. 
+
+By accessing or using GEO-Bench-2, you agree to comply with the original dataset licenses and acknowledge that the responsibility for verifying appropriate use lies with the end-user.