U of M expulsion over alleged AI cheating upheld on appeal

Listen to this article

A University of Minnesota student was expelled from a doctoral program in 2024 following allegations that he used artificial intelligence to cheat. In an opinion filed Feb. 2, the Minnesota Court of Appeals affirmed the university’s decision.

Haishan Yang was a third-year doctoral student in health economics at the University of Minnesota. In August 2024, Yang sat for an exam that permitted the use of class materials but prohibited using any artificial intelligence tools, such as ChatGPT.

A four-member faculty grading committee became concerned that Yang had used some sort of artificial intelligence software on the exam. One faculty member ran the exam questions through an AI program and saw comparability. Another discussed the accusations with Yang, eventually reporting the matter to the Office for Community Standards. The report included allegations that Yang violated the scholastic dishonesty provisions of the Student Code of Conduct by using AI to produce his answers on the exam, and it recommended Yang be dismissed from the program.

Yang sent a 35-page response denying the allegations. The office then suggested an informal resolution of expulsion or a formal hearing. Choosing the formal hearing, Yang was represented by an advocate from Student Advocate Services. Both parties presented exhibits, examined and cross-examined witnesses, and gave opening and closing statements.

Yang was informed that the hearing panel unanimously concluded that he had violated the code and more likely than not had used AI on the exam. Although the panel did not rely on AI-detection evidence, it credited graders’ ability to identify AI-written work, pointed to irrelevant sources that raised concerns about AI use, and cited Yang’s lack of citations, repeated excuses, and inconsistent testimony. Concluding that Yang’s conduct had destroyed the trust essential between doctoral students and professors and that it could not be restored, the panel ordered his expulsion.

In response, Yang filed several lawsuits between late December 2024 and March 2025. Yang first sued a participating professor in Hennepin County District Court on Dec. 26, 2024, then filed a federal lawsuit on Jan. 8, 2025, against several university employees involved in the hearing. On March 3, 2025, he petitioned the Minnesota Court of Appeals for a writ of certiorari challenging the university’s conduct decision, and on March 31, 2025, he filed another Hennepin County case against the university alleging violations of the Minnesota Government Data Practices Act. Yang also filed a complaint with the Minnesota Office of Administrative Hearings. Some of the lawsuits are open. In October 2025, a federal court dismissed Yang’s suit, finding that his due process rights were not violated.

In the case before the Court of Appeals, filed against the University of Minnesota, Yang argued that the university did not follow its own internal procedures, made unsubstantiated findings, and violated his due process rights. However, the court found in favor of the university.

Yang argued that the university violated its hearing procedures by allegedly interrupting him, limiting his cross-examination, imposing time limits, and admitting over his objections unreliable evidence, including AI-generated materials, a dismissed allegation, and an AI-probability score. He also challenged an off-the-record discussion that resulted in denying admission of new evidence related to AI. None of this, to the court, demonstrated that the university violated its internal procedures.

Additionally, Yang argued that the panel’s findings were not supported by substantial evidence. However, the court found that the hearing panel simply found the university’s evidence persuasive, credited graders’ ability to identify AI-generated work, and cited irrelevant documents, missing citations, repeated excuses, and inconsistencies in the relator’s testimony as signs of AI use. After reviewing the record, the court concluded that substantial evidence supported the panel’s finding that Yang likely used AI and violated the code.

Nor did the court find that the university violated Yang’s procedural or due process rights. Yang argued that his procedural due process rights were violated because the university did not provide a link to certain AI-generated evidence until the end of the hearing, allegedly depriving him of adequate notice and a meaningful opportunity to respond. The court noted that Yang had months of notice about the AI evidence and the charges against him and received a full evidentiary hearing where he could present and challenge evidence.

Yang also claimed that his substantive due process rights were violated when the university relied on mistrust to justify expulsion and by allowing allegedly biased graders and a biased hearing chair to influence the outcome. However, the court noted that an arbitrary or a substantial departure from accepted academic norms was needed to demonstrate a substantive due process violation. In this case, the court said, the panel’s decision was based on the seriousness of scholastic dishonesty and the importance of trust in the program, and expulsion was a permitted sanction under the code. The court also noted that Yang failed to show bias or lack of professional judgment by the actual decision-makers.