@article{auditinglanguagemodelsforhidden, title = {Auditing language models for hidden objectives}, author = {Samuel Marks and Johannes Treutlein and Trenton Bricken and Jack Lindsey and Jonathan Marcus and Siddharth Mishra-Sharma and Daniel Ziegler and Emmanuel Ameisen and Joshua Batson and Tim Belonax and Samuel R. Bowman and Shan Carter and Brian Chen and Hoagy Cunningham and Carson Denison and Florian Dietz and Satvik Golechha and Akbir Khan and Jan Kirchner and Jan Leike and Austin Meek and Kei Nishimura-Gasparian and Euan Ong and Christopher Olah and Adam Pearce and Fabien Roger and Jeanne Salle and Andy Shih and Meg Tong and Drake Thomas and Kelley Rivoire and Adam Jermyn and Monte MacDiarmid and Tom Henighan and Evan Hubinger}, year = {2025}, eprint = {2503.10965}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2503.10965v2}, }