Evaluating large language models for accuracy incentivizes hallucinations. (nature.com)

Authors: Kalai AT … Zhang E Journal: Nature Published: 2026-05-01 DOI: https://doi.org/10.1038/s41586-026-10549-w