The Procedural Minefield: Navigating the AI Integrity Crisis in Higher Education
It is the final week of the semester. A faculty member sits at their desk, reviewing a stack of student papers. One submission catches their eye—something feels "off." The citations are technically real, but two point to a textbook the class never used. The prose in the introduction is cogent and academic, yet the middle of the paper shifts into a disjointed, generic voice. Out of curiosity, the professor runs the text through an AI detector, which returns a high probability of machine generation.
In this moment, the professor faces a dilemma that has become the defining challenge of the current academic year. Do they email the student? Fail the assignment? File a formal academic-integrity report with the dean? Or simply lower the grade and hope the issue disappears?
While the field of higher education has spent the last three years debating whether the use of generative AI constitutes "cheating," a more pressing, practical, and legally perilous question has gone largely unanswered: What are faculty and administrators actually supposed to do when they suspect AI misuse?
The Legal Framework: Due Process and Academic Judgment
The uncertainty surrounding AI enforcement is not merely a matter of classroom policy; it is a significant legal liability for colleges and universities. The tension lies in the distinction between "academic judgment" and "disciplinary determinations"—a divide the U.S. Supreme Court has been policing for over half a century.
In the 1975 landmark case Goss v. Lopez, the Supreme Court established that public school students possess a constitutionally protected interest in their education, meaning the state cannot strip them of that interest without due process. Three years later, in Board of Curators v. Horowitz, the Court refined this, distinguishing between the professional evaluation of academic work and the formal punishment of misconduct.
Academic grading is largely shielded from legal challenge because it relies on the professional expertise of the professor. However, once an institution moves to label a student as a "cheater," it initiates a disciplinary process. This triggers procedural requirements: notice, the right to respond, and an impartial appeals process. When universities blur these lines—for instance, by allowing a professor to unilaterally mark a student’s record for "AI cheating" without a formal hearing—they invite litigation.
Recent case law illustrates the danger of procedural shortcuts. In Yang v. Neprash, a federal court upheld an expulsion for AI use because the University of Minnesota provided robust due process, including an evidentiary hearing and appellate review. Conversely, in Matter of Newby v. Adelphi University, a state court annulled an integrity finding because the university’s appeal process was "inconsequential"—the same administrator who issued the penalty also presided over the appeal. The lesson is clear: courts do not necessarily care if a student used AI; they care deeply about whether the university followed its own rules.
A Step-by-Step Decision Framework for Faculty
To navigate this landscape, faculty need more than instinct. They need a systematic framework that separates academic evaluation from disciplinary action.
Step 1: Distinguishing Evidence from Hunch
The first mistake many faculty make is relying on AI detectors. These tools are scientifically contested, prone to false positives, and frequently banned by institutional policy as the sole basis for discipline. A detector score is not evidence; it is a suggestion.
Instead, faculty should focus on "hard" versus "soft" indicators. Hard indicators—such as citations to non-existent articles, references to textbooks not used in the course, or content that contradicts in-class observation—are objective, verifiable facts. Soft indicators, such as the frequent use of em dashes, repetitive hedging (e.g., "it is important to note"), or a sudden shift in stylistic register, are helpful for identifying where to look, but they are insufficient for a formal charge.
Step 2: Syllabus Transparency
Courts and internal review boards prioritize "notice." If a syllabus does not explicitly state the parameters of AI usage, a professor is on shaky ground. Vague phrases like "all work must be original" or "adhere to academic integrity" are often insufficient to inform a student that a specific, nuanced use of AI is prohibited. Fairness requires that the student knows the rules before the assignment is submitted.
Step 3: Grading vs. Misconduct
This is the most critical juncture. If a student fails to meet the criteria of an assignment—for example, if a paper is poorly written or fails to engage with course material—that is an academic judgment. The professor grades the work accordingly, perhaps giving it an ‘F’ or a ‘zero.’
If the student is accused of misconduct—an intentional violation of the university’s integrity code—that is a different matter. Faculty cannot simply relabel a low grade as a "cheating violation" to bypass the headache of an integrity hearing. If the goal is a transcript notation or disciplinary record, the institution’s formal misconduct process must be followed.
Step 4: Procedural Calibrations
The level of process must match the severity of the consequence. If a professor wants to offer a student a chance to rewrite an assignment, a brief conversation suffices. If the penalty involves a failing grade for the entire course or expulsion, the student is owed a formal, documented process. Following this procedure is the best protection for both the institution and the professor against future lawsuits.
Institutional Implications and Responsibilities
The current "AI moment" has exposed a lack of institutional infrastructure. Many universities have released statements declaring AI use a violation of policy, yet they have failed to provide the necessary support for faculty to enforce those policies fairly.
To mitigate risk, institutions must prioritize five areas:
- Clear Definitions: Explicitly defining what constitutes "unauthorized use" of generative AI.
- Evidentiary Standards: Establishing whether a preponderance of evidence or "clear and convincing" evidence is required for AI-related sanctions.
- Training for Adjudicators: Ensuring that those who hear appeals are not the same individuals who initially levied the accusation.
- Detector Policy: Providing official guidance on whether (and how) AI detection software can be used.
- Appeals Mechanisms: Ensuring that the appeals process is robust, independent, and transparent.
The Human Element
The faculty member in our opening scenario now has a clearer path. The "hard evidence" of the phantom citations justifies a conversation or a referral to the academic integrity office. However, the AI detector score should remain in the background—a starting point for investigation, not a smoking gun for expulsion.
By distinguishing between the act of grading and the act of charging a student with a violation, faculty can avoid the most common legal pitfalls. When a professor uses their syllabus to set clear expectations and relies on verifiable, hard evidence rather than algorithmic hunches, they protect their own professional autonomy.
Ultimately, the debate over AI is not going to vanish. Students will continue to use these tools, and faculty will continue to be suspicious of the results. The goal of the university should not be to eradicate the use of technology, but to create a system that is fair, predictable, and legally sound. Institutions that refuse to codify their procedures will find themselves increasingly vulnerable, forced to answer to judges for decisions that should have been resolved with better institutional policy.
In the absence of clear guidance, faculty are forced to improvise. But in the realm of due process, improvisation is the precursor to litigation. The time for universities to finalize their procedures is not next semester; it is now.
Christian Moriarty is a professor of ethics and law at St. Petersburg College, director and treasurer for the International Center for Academic Integrity, and ethics and governance lead for the Florida AI Learning Consortium. This article reflects his view of the current state of American law as it bears on academic-integrity proceedings and does not constitute formal legal advice.
