Dartmouth provost resigns after faculty committee finds undisclosed AI use

Dartmouth President Sian Leah Beilock announced on October 28, 2026 that Provost Santiago Schnell had resigned as provost and was leaving the university. Beilock made the announcement via campus email. Matthew Delmont, a history professor who holds the faculty post of associate dean for international and interdisciplinary studies, will be interim provost through June 2028.

The resignation followed a faculty-led committee’s determination that Schnell used artificial intelligence in his writing without proper disclosure. The committee’s report, dated October 27, 2026, found Schnell changed his writing significantly using AI in violation of at least one publication’s disclosure requirement. It also determined that Schnell had not been open about his personal AI use and had not met Dartmouth’s ethics and professionalism standards.

Schnell provided his Dartmouth-issued laptop but did not provide logs from his personal large language model (LLM) accounts or make himself available to meet with the committee. Ars Technica reported that the report described Schnell’s cooperation with the investigation as insufficient. The committee did not use AI detection tools, relying instead on Dartmouth enterprise Claude logs and e-mails, and it recommended no disciplinary action.

Beilock formed the committee on September 25, 2026 after news reports questioned Schnell’s AI use, including an August 2026 Semafor article about AI-written opinion pieces naming his Washington Post column. Schnell was named Dartmouth’s chief academic officer at the start of the 2025-26 academic year, joined in July 2025, and had published at least 30 pieces by the time of the report.

Schnell, a mathematical biologist and native of Venezuela, served as dean of the College of Science at the University of Notre Dame and addressed the U.S. Conference of Catholic Bishops plenary assembly in Orlando, Florida, in June. He wrote three op-eds for the National Catholic Register this year, two on AI and education, and The Dartmouth’s report on his AI usage flagged all three. Schnell wrote that large language models used with discipline can be genuinely useful and that AI offers an opportunity for classrooms to return to low-tech, personalized learning.

On August 9, 2026, Schnell wrote a Washington Post op-ed titled “Universities are fighting AI cheating. But there’s a deeper problem.” An August 2026 Semafor investigation found the AI detector Pangram flagged Schnell’s Washington Post editorial as at least 80 percent AI-generated. The Dartmouth reported that Pangram rated 96 percent of Schnell’s nine professional articles that year as AI-written.

Experts consider AI detection tools such as Pangram unreliable, and Schnell said they are biased against nonnative English speakers.

The committee found that for a ChemSystemsChem paper, Schnell used Claude to edit his manuscript, fix code errors, and create figures after reviewers asked for revisions. Claude enterprise logs showed Schnell used the LLM to revise papers, find and fix code bugs, and edit or generate figures. A single Claude prompt instructed the model to review a ChemSystemsChem submission as an expert in mathematical chemical kinetics and also in nonlinear chemical dynamics. Schnell uploaded the manuscript of a forthcoming article to Claude and asked whether it showed signatures of AI writing, then later submitted a revised version of that article incorporating changes Claude suggested.

Schnell acknowledged using AI for language refinement, copy editing, clarity and organization, but said it did not substitute for his research, intellectual judgment or authorship. In remarks to Semafor, Schnell said he used chatbots to sharpen a handful of arguments and check grammar mistakes. He told Semafor, “to refine a few arguments” and “check for grammatical errors.”

Schnell said he relied on older author instructions and understood disclosure to begin in 2025, while the committee found it began in 2023. The journal required disclosure of that AI use and Schnell did not disclose it; the publisher had required AI disclosure since at least 2023. He told a journal editor he used AI editorially and did not disclose it because he did not understand AI-assisted editing required disclosure.

After the uproar, Schnell asked the editor to pause publication and disclosed he used AI as an editorial tool to refine sentences and arguments; the committee said that statement understated his actual AI use. Students and faculty argued Schnell violated the college honor code, which requires students to acknowledge generative AI use in coursework, and said he should lose his job. Schnell resigned less than two weeks before faculty leaders planned to hold a no-confidence vote against him.

David Kotz, a former provost who teaches computer science, led the committee. Members included English professor Peter Orner, linguistics professor Lindsay Whaley, trustee Odette Harris, and computer science professor Hany Farid. The committee found no violation in a fourth paper co-authored by Schnell.

Andrew Miltenberg, Schnell’s attorney, said on October 30, 2026 that Schnell resigned out of respect for colleagues and did not get a meaningful chance to share his side during the internal review. “Dr. Schnell has decided to resign as Provost and from the Dartmouth faculty.” Miltenberg said. Some scholars questioned whether Schnell faced extra scrutiny and said acceptable AI norms in higher education remain largely undefined.

Beilock said senior leaders must show the transparency and human judgment expected from students and scholars by the university. Dartmouth’s honor code requires students to acknowledge generative AI use in coursework, and students and faculty argued Schnell violated that code.

Dartmouth encouraged AI adoption through an agreement with Anthropic and Amazon Web Services and a conference called AI at 70: What Must Remain Human?. The college entered an agreement last year with Anthropic and Amazon Web Services to provide faculty and students access to AI technology. Dartmouth calls itself the birthplace of AI, and Beilock has called for more AI at the university.