@article{loganalysisisnecessaryforcredibleevaluat, title = {Log analysis is necessary for credible evaluation of AI agents}, author = {Peter Kirgis and Sayash Kapoor and Stephan Rabanser and Nitya Nadgir and Cozmin Ududec and Magda Dubois and JJ Allaire and Conrad Stosz and Marius Hobbhahn and Jacob Steinhardt and Arvind Narayanan}, year = {2026}, eprint = {2605.08545}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2605.08545}, }