Dartmouth provost Santiago Schnell resigns over undisclosed AI use

The case began with press scrutiny rather than an internal audit. President Sian Leah Beilock formed the committee on Sept. 25 after news reports questioned Schnell's AI use, including an August Semafor article about AI-written opinion pieces that named his Washington Post column. That sequence matters: external reporting triggered the review, and the committee's report is dated Oct. 27, roughly a month later.

The committee's method is notable. It explicitly declined to use AI detection tools, relying instead on Schnell's logs from Dartmouth's enterprise Claude deployment plus his emails — an approach that turns workplace tool records into the evidentiary basis. The report also says Schnell did not fully cooperate, noting he provided his Dartmouth-issued laptop but not logs of personal LLM accounts used for academic and public writings.

The findings span several publications. In one journal paper, the committee found he used Claude to substantially edit a research manuscript, troubleshoot code errors and generate figures without the required disclosure; the publisher had required AI disclosure since at least 2023. In a forthcoming article, he asked Claude whether there were "many signatures of AI writing" and submitted a version with the suggested changes. The committee found no violation in a fourth, co-authored paper. Schnell has published at least 30 pieces since joining Dartmouth in July 2025, so the committee focused on a handful given the short timeline. Beilock framed the outcome around leaders modeling "the transparency and human judgment that we expect from our students and scholars" — a standard set days before the university's own "AI at 70: What Must Remain Human?" conference.