Story
Oct 31, 2026
Dartmouth’s AI Champion Resigns Over the Transparency He Preached
Dartmouth’s leadership argues that a university can push aggressively into AI without abandoning human judgment, but Schnell’s resignation shows how quickly that promise collapses when disclosure rules are treated as an afterthought.
Dartmouth’s AI debate sharpened in August, when Provost Santiago Schnell published a Washington Post essay warning that universities faced a problem deeper than ordinary AI cheating. Behind that argument was his own expansive view of the technology: AI could help develop, edit and test ideas, while a human author remained responsible for the final work.1
That distinction became consequential after Dartmouth began examining Schnell’s use of AI in his research and public writing. A faculty committee reviewed logs from the university’s Claude enterprise account and found he had used the tool to substantially edit a ChemSystemsChem manuscript, troubleshoot code and generate figures. The journal required disclosure of that kind of assistance, the committee said, but Schnell did not provide it.2
The review also focused on Schnell’s August Post essay. He had told Semafor he used ChatGPT to “refine a few arguments” and “check for grammatical errors,” but Dartmouth concluded the newspaper’s standards required that assistance to be disclosed as well.2 The committee further found that Schnell had not been forthcoming and had fallen short of institutional ethical standards.2
Schnell’s defenders could point to an important nuance: the available logs suggest he used AI not simply to generate finished prose, but as an editorial and technical collaborator. Yet the committee said the full extent of that use remained unclear, in part because he did not provide personal-account logs or meet with investigators.1
By Wednesday, Schnell had resigned, becoming a rare high-profile academic leader to leave over AI-disclosure failures. President Sian Leah Beilock framed the outcome not as a retreat from Dartmouth’s AI ambitions, but as a test of leadership. “It is so critical that we model the transparency and human judgment that we expect from our students and scholars,” she wrote.2
That is the central contradiction: Dartmouth still wants more AI, but its provost’s departure makes clear that the institution’s tolerance for experimentation ends where undisclosed assistance begins.