1 paper
Kevin Wei, Sumaya Nur Adan, Stephan Llerena +17
We introduce FraudBench, an evaluation of large language models' (LLMs) propensities to recommend, assist, or directly take actions that would constitute fraud under U.S. law. Our…