Dan Hendrycks

Executive and Research Director

Center for AI Safety

Areas of expertise: Adversarial robustness, AI alignment, AI regulation, Biosecurity, Corporate governance, Democratic AI, Explaining AI risk, Extinction risk scenarios, International competition, Misuse, ML community, Nuclear, Related risks

Dan received his PhD from UC Berkeley where he was advised by Dawn Song and Jacob Steinhardt. He is now the director of the Center for AI Safety. He received his BS from UChicago. His research is supported by the NSF GRFP and the Open Philanthropy AI Fellowship. Dan has had a long-standing interest in AI safety research and has also contributed to advance AI more generally. He helped contribute the GELU activation function (the most-used activation in state-of-the-art models including BERT, GPT, Vision Transformers, etc.), the out-of-distribution detection baseline, and distribution shift benchmarks.

Selected Publications

Hendrycks, D. (2023). Natural Selection Favors AIs over Humans. arXiv preprint

Hendrycks, D., Mazeika, M., & Woodside, T. (2023). An Overview of Catastrophic AI Risks. arXiv preprint