@article{aicontrolimprovingsafetydespite, title = {AI Control: Improving Safety Despite Intentional Subversion}, author = {Ryan Greenblatt and Buck Shlegeris and Kshitij Sachan and Fabien Roger}, year = {2023}, eprint = {2312.06942}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2312.06942v5}, }