Dartmouth Provost Resigns After Review Finds Undisclosed AI Use in Published Writing

Dartmouth Provost Resigns After Review Finds Undisclosed AI Use in Published Writing

A senior academic leader at Dartmouth College in the United States has resigned following an investigation into his use of artificial intelligence in academic publications and public writing.

Santiago Schnell, who became Dartmouth's provost in July 2025, announced his departure after an independent, faculty-led committee reviewed concerns about his use of AI tools, including ChatGPT and Claude.

On 28 October 2026, Dartmouth President Sian Leah Beilock confirmed that Schnell had resigned as provost and would leave the university. Beilock described the departure as Schnell's decision and said the institution remained grateful for his leadership.

The committee concluded that Schnell had used AI to meaningfully alter published writing without meeting disclosure requirements in at least two instances. It also found that he had not been sufficiently forthcoming about his use of the technology and that his conduct fell short of Dartmouth's standards for professional and academic integrity.

The case has drawn wider attention because universities are still developing policies on when AI can be used in research, writing and teaching, and when its use must be disclosed.

What triggered the investigation?

Concerns about Schnell's writing became public in August and September 2026 after reports questioned the extent of AI assistance in his published work.

One of the articles under scrutiny was an opinion piece he wrote for The Washington Post in August, titled Universities are fighting AI cheating. But there's a deeper problem.

The article discussed the challenges universities face as students increasingly use AI tools in their academic work.

An investigation by Semafor reported that the article had been flagged as AI-written by the detection tool Pangram. Subsequent reporting by Dartmouth's student newspaper examined other recent publications by Schnell and raised further questions about how extensively he had used AI.

Schnell acknowledged using ChatGPT to refine arguments and check grammar, while maintaining that the underlying ideas were his own.

In September, he also wrote to Dartmouth's community acknowledging that he should have explicitly disclosed AI use in his public writing and paid closer attention to publishers' requirements.

Dartmouth President Beilock subsequently appointed an independent committee to examine his published work, use of AI and compliance with university and publication policies.

An important distinction: The committee did not rely on AI-detection scores to reach its conclusions. Its investigation examined available records, communications, publication requirements and chat histories from Dartmouth's enterprise Claude system.

What did the committee find?

The committee reviewed selected publications Schnell produced after joining Dartmouth in July 2025. Its report, dated 27 October 2026, examined whether he had complied with publication requirements, accurately disclosed his AI use and followed Dartmouth's ethical standards.

The investigation identified concerns involving several pieces of writing.

1. An academic research paper

One publication examined was Reaction Acceleration: Reviving the second derivative in chemical kinetics, published in the journal ChemSystemsChem in September 2026.

According to the committee, Schnell used Dartmouth's enterprise version of Claude to substantially edit the research manuscript, troubleshoot computer-code errors and generate figures after receiving major revisions from peer reviewers.

The journal's applicable policies required disclosure of AI use of this kind, but Schnell had not disclosed that assistance when submitting or publishing the paper.

The committee also examined his explanation that he had relied on older author instructions. It concluded that the publisher had required AI-use disclosure since at least 2023.

Schnell later informed the journal about his use of ChatGPT, describing assistance with language, code review, reproducibility documentation and consistency checks. The committee said that disclosure did not include his use of Claude.

2. A Washington Post opinion article

The committee also reviewed Schnell's August 2026 opinion article about AI and academic cheating.

Schnell had told Semafor that he used ChatGPT to refine arguments and check grammar before submitting the article.

However, The Washington Post required authors to confirm that their submissions had not been created or manipulated using artificial intelligence or editing software.

The committee therefore concluded that his AI use violated the newspaper's stated policy because it had not been disclosed and did not comply with the rules for submitted work.

3. Another academic paper

The committee reviewed a separate paper published in the Bulletin of Mathematical Biology, co-authored by Schnell and two other researchers.

The report said that Schnell had used AI coding tools to optimise and review computer code associated with the research.

However, the committee found no evidence that he had used AI to edit or manipulate the paper's text itself. It also concluded that the publisher's policy did not clearly prohibit the use of AI for this type of code work or clearly require disclosure under the circumstances reviewed.

The committee therefore did not conclude that this paper violated the publication's policy.

This distinction matters: the investigation did not find that every publication it reviewed involved a confirmed policy violation.

Questions about transparency and cooperation

The committee's concerns extended beyond the use of AI itself.

It said Schnell had not provided all the information requested during the investigation. Although he made his Dartmouth-issued laptop available for inspection, he did not provide logs from personal AI accounts used for his academic and public writing, despite the committee's request.

The report also said he failed to make himself available at the requested time or propose alternative times to discuss his use of AI.

The committee concluded that some of his disclosures and explanations did not fully reflect the extent of his AI use.

It also examined his handling of an article intended for the Journal of Catholic Higher Education. According to the report, Schnell used Claude during the manuscript's preparation and revision process, and the committee believed that his description of this assistance understated what the available records showed.

The committee said it had to limit its review because of the time available and the information it could obtain. It therefore examined selected publications rather than attempting to assess everything Schnell had written.

Why did he resign?

Dartmouth's president said the committee found that Schnell's conduct fell short of standards set out in the university's Code of Ethical Conduct, Academic Honor Principle and relevant professional policy at its medical school.

These standards emphasise honesty, integrity, transparency and responsibility in academic and professional work.

The provost is the university's chief academic officer and occupies a senior leadership position. The committee said such a leader should model the academic integrity expected of students, faculty and other members of the institution.

After receiving the findings, Dartmouth announced that Schnell had resigned as provost and was leaving the university.

The president stressed that the departure was Schnell's decision.

The committee's report did not recommend related disciplinary action under the procedures for assessing complaints under those policies. Nevertheless, its findings identified conduct that it considered inconsistent with the institution's standards.

Who will replace him?

Dartmouth announced that Matthew Delmont, a professor of history and associate dean of faculty for international and interdisciplinary studies, would serve as interim provost through June 2028.

Delmont will work with President Beilock, university deans and the provost's office to support faculty, staff and students during the transition.

The university has not described the interim appointment as a permanent replacement.

The wider debate over AI and academic integrity

The controversy comes at a time when universities, publishers and researchers are debating how AI should be used in academic work.

AI tools can help researchers review code, improve clarity, organise information, analyse data and prepare documents. However, institutions and publishers may impose different rules depending on the purpose of the work and the role played by the technology.

Some publishers allow limited language editing but require disclosure when AI contributes more substantially to the manuscript. Others prohibit AI-generated or AI-manipulated submissions under specific conditions.

This creates an important responsibility for researchers: they must understand the rules that apply to each publication and follow them accurately.

The Dartmouth case illustrates that the question is not simply whether AI was used. It is also whether its use was permitted, whether it changed the substance or presentation of the work, whether disclosure was required and whether the author described that use honestly.

A tool's ability to generate or improve text does not remove the author's responsibility for the final work.

What does this mean for students and researchers?

The case offers several practical lessons for anyone using AI in academic or professional writing.

Read the rules before submitting work. A tool that is permitted for one assignment or publication may be restricted by another.

Disclose AI use when required. Authors should follow the specific disclosure requirements set by their university, journal or publisher.

Remain responsible for the final result. Researchers and writers must check the accuracy, integrity and quality of material prepared with AI assistance.

Distinguish assistance from authorship. Correcting grammar, reviewing code, generating figures and rewriting substantial sections can involve different levels of AI contribution. Authors should understand how the applicable rules treat each activity.

Keep appropriate records. Clear records of AI assistance can help researchers explain how they used the technology and demonstrate compliance with publication requirements.

These lessons apply to students, lecturers, researchers, journalists and professionals who use AI to prepare work that will be evaluated or published.

A major test for universities adopting AI

Dartmouth presents itself as a university engaged in the development and discussion of artificial intelligence. The controversy surrounding its former provost has therefore raised particular questions about how institutions should model responsible use of the technology.

The university's response emphasised that AI policies must be accompanied by transparency, human judgement and consistent standards.

The case does not establish that all AI-assisted writing is unethical or prohibited. Rather, it demonstrates the consequences that can follow when a person's use of AI conflicts with a publication's rules and the person fails to disclose that use appropriately.

As AI becomes more common in research and education, universities will need policies that explain acceptable use while ensuring that academic honesty and accountability remain central.

Source: Dartmouth College's Office of the President, the student newspaper The Dartmouth, and reporting by The Washington Post and the Associated Press.

Official university statement: https://president.dartmouth.edu/news/2026/10/update-provost-review

Full committee report: https://president.dartmouth.edu/sites/president/files/2026-10/Advisory%20Committee%20Report%20to%20President%202026-10-27.pdf

Additional reporting: https://www.thedartmouth.com/article/2026/10/weaverbell-klee-sharma-santiago-schnell-resigns-as-provost-of-the-college