Santiago Schnell arrived at Dartmouth’s provost office in July 2025 as the university wrestled with a problem it helped create: how to lead on artificial intelligence without surrendering academic standards. Dartmouth celebrates its 1956 role in launching the field, while students and faculty confront a rapidly blurring line between assistance and authorship.¹

In August 2026, Schnell published a Washington Post opinion essay warning that universities faced a deeper problem than AI cheating. He later told Semafor that he had used chatbots to “refine a few arguments” and check grammar—a description consistent with his broader view that AI could serve as an editorial tool rather than a co-author.²

But Dartmouth’s review of his enterprise Claude records found more extensive use in a ChemSystemsChem manuscript: substantial editing, code troubleshooting and figure generation. The journal required disclosure of that kind of assistance, the committee concluded, and Schnell had not provided it. Investigators also said he was insufficiently forthcoming, declining to provide personal-account logs or meet with the committee, leaving the full scope of his use unresolved.¹

The clash was not over whether scholars may use AI at all. Schnell had argued that human authors remain responsible for work shaped with computational help. Dartmouth President Sian Leah Beilock, who has pushed to keep the college at the forefront of AI, framed the obligation more sharply after the review: leaders must model “the transparency and human judgment that we expect from our students and scholars.”²

On Wednesday, Schnell resigned. Beilock said the departure was his decision and named historian Matthew Delmont interim provost through June 2028. The episode leaves Dartmouth pursuing AI ambition under a harder lesson: in academia’s new mixed-authorship era, disclosure is not a footnote—it is the test of trust.²