A Yale AI cheating lawsuit has turned one Executive MBA final exam into a 13-count federal case over academic discipline, AI detection, political bias and one very belated Apple Pages file.
Thierry Rignol sued Yale University in February 2025 after the school suspended him for a year and gave him an F in Sourcing and Managing Funds, a course in its Executive MBA program. Rignol says he paid Yale $208,500 in tuition and was on track to finish first in his class before the discipline cost him a shot at valedictorian.
According to Rignol’s complaint, Yale punished him after his final exam was flagged for possible use of generative AI. One professor ran the text through GPTZero, an AI-detection service, which reported that large portions were likely machine-written. Rignol argues those tools are unreliable and biased against non-native English speakers, including him. He says his polished, structured writing reflected his academic record and French background, not cheating.
Yale’s filings describe a broader set of concerns. K. Geert Rouwenhorst, one of the course professors, told a dean in June 2024 that GPTZero was only part of the issue. He also cited overlap between one answer and ChatGPT output responding to the same prompt, weaker performance on a question where AI would have been less useful, and doubts about whether a student could produce such a long and clean exam in four hours.
What is the Yale AI cheating lawsuit about?
The dispute began with a spring 2024 final exam in MGT423E, Sourcing and Managing Funds. The exam was open book, closed Internet and barred AI tools. Students wrote answers on their computers, converted the work to PDF and submitted the PDFs. Among 72 students, Yale says only Rignol’s submission drew suspicion from a teaching assistant.
Rignol denies using AI and says Yale’s disciplinary process became a “sham” aimed at punishing his conservative political speech, including views he expressed in class about smaller government, pro-business policy and skepticism of DEI. His current complaint brings 13 causes of action, including breach of contract, civil rights violations, emotional distress, unfair trade practices, defamation and invasion of privacy.
He seeks damages tied to his health, reputation, earnings and career prospects. He also wants Yale to erase the disciplinary record, restore his grade and readmit him, though Yale says he already returned from suspension and graduated.
Generative AI systems such as ChatGPT produce text in response to prompts, which is why schools have tried to police their use on exams. The technical problem is that detection tools can produce false positives, a limitation Yale itself has recognized as making AI policing through detectors difficult. For more background, see Kernel’s explainer on how LLMs work.
Why did the exam file become central?
Yale says Rignol repeatedly failed to provide the underlying file used to create his submitted PDF. Finance professor James Choi, who led the Honor Committee review, emailed Rignol several times in August 2024 asking for the Microsoft Word file behind the exam submission and warning that a lack of candor could itself violate the honor code.
Rignol did not provide a Word file. At a November 2024 hearing, he said he had not used Microsoft Word at all. He had written the exam in Apple Pages. Yale’s lawyers later argued that this answer was evasive because Rignol could have corrected the request earlier and produced the relevant Pages file.
After the hearing, assistant dean Wendy Tsung asked Rignol for the Pages file, which he provided that afternoon. Yale then asked him to return to campus so the committee could inspect his laptop. Rignol said he could not come back on short notice. That evening, the Honor Committee found him responsible for not being forthcoming. Later in November, it also found he violated the exam rules and imposed the F.
The case remains in federal court. In June 2026, the judge allowed Rignol to file a third amended complaint but warned that another amendment would not be granted without extraordinary circumstances. Yale moved on July 15, 2026, to dismiss the expanded complaint, calling it conclusory and unsupported by well-pleaded facts.
This story draws on original reporting from Ars Technica.
