Dartmouth’s AI Transparency Champion Falls Over His Own Disclosures
Dartmouth’s AI Transparency Champion Falls Over His Own Disclosures
In August 2026, Dartmouth Provost Santiago Schnell entered the AI debate publicly with a Washington Post essay warning that universities faced a problem deeper than simple cheating. Yet his own use of the technology soon became the subject of an internal review.
The faculty committee examined logs from Dartmouth’s Claude enterprise account and concluded that Schnell had used the tool to substantially edit a research manuscript, troubleshoot code and generate figures for a ChemSystemsChem submission. It also found he had not disclosed AI assistance to the journal or to The Post, despite their requirements. The review said Schnell was not forthcoming and had fallen short of the college’s ethical standards.
Schnell’s reported explanation was narrower than the committee’s concern. He had described chatbots as a way to “refine a few arguments” and “check for grammatical errors,” framing them as an editorial aid rather than a substitute author. But the committee said the available evidence suggested his use went beyond those public descriptions; it also cited his failure to provide personal LLM logs or meet with investigators.
By September, President Sian Leah Beilock had acknowledged the difficulty of setting rules in a fast-moving field, saying that leading on AI would be “messy” and that institutions would make missteps. The inquiry nevertheless moved ahead at a university that celebrates its foundational role in AI research.
On Wednesday, Dartmouth announced Schnell’s resignation. Beilock said the decision was his and named history professor Matthew Delmont interim provost through June 2028. Her broader message drew the line Dartmouth now wants to enforce: people may differ over when to use AI, but leaders must “model the transparency and human judgment” expected of students and scholars.
The resignation leaves Dartmouth with an uncomfortable test case. Schnell’s position—that AI can assist a responsible human author—has not disappeared. But the university’s response makes clear that, for now, undisclosed assistance is the breach that matters most.
Write a comment