AI Cheating Lawsuits Tracker — Every Case, Who Won (2026)

AI Cheating Lawsuits Tracker: Every Case and Who Won (2026)

Students accused of using AI by a detector are starting to fight back in court. The coverage is scattered across one-off news stories—and the few "trackers" that exist get basic facts wrong (one popular aggregator labels its whole table "federal cases" when the most famous student win was a state proceeding).

This is a neutral, primary-source tracker of every US lawsuit we can verify where a student was accused of AI misconduct off the back of a detector. We label each by court, posture, and outcome, with docket numbers, and we update it as cases move. For the argument about what the pattern means, see the companion piece: why students who win these cases win on due process, not on the detector being debunked.

The cases at a glance

Scoreboard so far: one clean student win, two school wins, three pending. No court has yet ruled that an AI detector is itself unlawful—every decision has turned on process (was there a fair hearing? was contrary evidence considered?), not on the technology.

The decided cases

Newby v. Adelphi University — student won (and it was not federal)

This is the case everyone cites, and the one everyone mislabels. Freshman Orion Newby, who is autistic and enrolled in Adelphi's "Bridges" support program, had a World Civilizations essay flagged by Turnitin as 100% AI-generated. Two other tools (Grammarly and ZeroGPT—note: not GPTZero) called it human-written.

A judge annulled the misconduct finding and ordered Newby's record expunged, calling the university's determination "without valid basis and devoid of reason."

The crucial detail almost every outlet gets wrong: this was a New York State Supreme Court matter—a CPLR Article 78 proceeding in Nassau County (Matter of Newby v. Adelphi Univ., 2026 NY Slip Op 26021), decided January 2026. It was not a federal lawsuit, and it was not a damages judgment—it was an administrative-review challenge to a broken disciplinary process. That distinction matters for anyone trying to understand what these rulings actually establish.

Harris (RNH) v. Hingham — the actual federal ruling, and the school won

The one academic-integrity AI case that genuinely is federal cuts the other way. In Harris v. Hingham (U.S. District Court, District of Massachusetts, No. 1:24-cv-12437), parents of a Massachusetts high-school student sued after he was disciplined for using AI on an AP U.S. History project. In November 2024, the court denied their request to raise his grade and expunge the detention, finding the school "reasonably concluded" a policy violation. The plaintiffs signaled the case would continue; we have not confirmed a final disposition and will update when one is on the record.

Yang v. University of Minnesota — student lost on both tracks

PhD student Haishan Yang, a Chinese national, was expelled in 2025 after being accused of using AI on a preliminary exam (flagged via GPTZero plus faculty judgment). He lost on both fronts: the Minnesota Court of Appeals affirmed the expulsion in February 2026, and a parallel federal suit (Yang v. Neprash, D. Minn.) was dismissed without prejudice in October 2025. It's the clearest example of a court upholding AI-related discipline when the school followed its own process.

The pending cases

Rignol v. Yale — the Ivy League ESL case

Yale School of Management executive-MBA student Thierry Rignol (a French national and non-native English speaker—originally filed as "John Doe," but the court denied him anonymity) was suspended after GPTZero flagged a final exam. He alleges national-origin bias and breach of contract, and famously ran Yale's own president's writing through GPTZero to argue the tool is unreliable. A preliminary injunction was denied in May 2025; the case is ongoing (D. Conn., 3:25-cv-00159).

Doe v. University of Michigan — the disability-bias theory

A Michigan undergraduate ("Jane Doe") with documented anxiety and OCD alleges her formal, meticulous writing style in "Great Books 191" was misread as AI. Notably, no commercial detector or score is named—the accusation rested on the instructor's judgment and what the complaint calls a "circular methodology" (prompting an AI with the student's own outline, then treating the similarity as proof). The theory is disability discrimination under the ADA/Section 504. The court denied a preliminary injunction in May 2026; a motion to dismiss is pending (E.D. Mich., 2:26-cv-10451).

Kato v. Palo Alto USD — the civil-rights statistics theory

The newest and most aggressive theory. A parent, Takashi Kato, sued after his son's essay on The Crucible was flagged 76% AI by Turnitin (N.D. Cal., filed May 2026). The complaint alleges Title IX and Title VI discrimination with statistical claims (that boys in the class were flagged far more often than girls), and that administrators re-ran the student's handwritten rewrite through Turnitin without the family's consent. It's a K-12 case, but the discrimination-by-detector framing imports directly to higher ed. As a heads-up for accuracy: the consent/privacy issue is pleaded as an allegation, not a standalone FERPA cause of action (FERPA has no private right of action).

Accused, cleared, but never sued

Not every false accusation becomes a lawsuit—many are resolved internally, which is why a tracker is narrower than the headlines suggest. Two widely-cited examples:

• William Quarterman (UC Davis, 2023) — flagged by GPTZero, cleared his name with Google Docs edit history (and by showing GPTZero flagged the Declaration of Independence and the Bible). No suit filed.
• University at Buffalo (2025) — a falsely-flagged grad student sparked an 1,100+ signature petition to disable Turnitin's AI detector. Protest, not litigation.

For more of these first-person stories, see our collection of students falsely accused by AI detectors—and if you're feeling the dread of being wrongly flagged, that's flagxiety, and it's common.

The backdrop: detectors under pressure

Two developments outside the courtroom shape these cases:

• Regulators are skeptical. In FTC v. Workado (final order August 2025), the Federal Trade Commission found an AI detector that advertised 98% accuracy actually performed at about 53%—"essentially a coin toss"—and barred the unsupported claims. A parallel market of AI "humanizers" that promise to beat detectors trades on equally shaky numbers.
• Universities are switching detectors off. Vanderbilt (2023), the University of Waterloo (2025), Michigan State, and Yale have disabled or deprioritized AI detection, often citing false positives and bias against non-native English writers. Michigan State's guidance is the cleanest statement of where institutions are landing: detector outputs are "potential indicators—not conclusive evidence" and "should never serve as the sole basis" for a finding. And testing bodies are going further still—the LSAT is ending remote testing and moving back in-person over integrity concerns.

The peer-reviewed evidence behind that retreat—including a Stanford study finding detectors flagged 61% of non-native English essays as AI—is covered in our breakdowns of AI-detector false-positive rates and why detection tools are biased against international students. For which schools actually run these tools, see do colleges use AI detectors, and on Turnitin's own accuracy problems, Turnitin's 15% miss rate.

What the cases establish (so far)

Put plainly: students who win, win on due process—a school ignoring exculpatory evidence, denying a meaningful appeal, or refusing a disability accommodation. Schools that win, win on proper process—pairing a detector flag with human review and a fair hearing. No court has yet ruled the detectors themselves invalid. We unpack that pattern, and why the popular "first federal AI win" framing is exactly backwards, in the companion analysis.

---

Sources and methodology

We include only US cases we can tie to a court record or reputable reporting, and we label each by court (state vs. federal), docket, posture (injunction, motion to dismiss, final ruling), and outcome. Primary and best sources:

• Matter of Newby v. Adelphi Univ., 2026 NY Slip Op 26021 (N.Y. Sup. Ct., Nassau County) — via Justia and New York Official Reports.
• Harris v. Town of Hingham, No. 1:24-cv-12437 (D. Mass.) — court docket and contemporaneous reporting.
• Yang v. Neprash (D. Minn.) and the Minnesota Court of Appeals decision — CourtListener and legal-press analysis.
• Rignol v. Yale University, 3:25-cv-00159 (D. Conn.) and Doe v. Univ. of Michigan, 2:26-cv-10451 (E.D. Mich.) — federal dockets and coverage from GovTech, EdScoop, and Inside Higher Ed.
• Kato v. Palo Alto USD, 5:26-cv-04078 (N.D. Cal.) — complaint and reporting from the SF Standard and Palo Alto Online.
• FTC v. Workado, LLC — Federal Trade Commission complaint and consent order.

Where a case's final disposition is unconfirmed (notably Hingham), we say so. This page is updated as cases progress; flagged outcomes are not predictions.

GradPilot helps students write and review essays in their own voice—so the question of "did a tool write this?" never comes up.

Related Reading

• AI Cheating Lawsuit Myths — Why Students Win on Due Process
• Flagxiety Stories: Students Falsely Accused by AI Detectors
• AI Detector False-Positive Rates Compared
• AI Detection Tools Are Biased Against International Students
• Do Colleges Use AI Detectors? The Truth About Turnitin