1. Press Release Draft
FOR REVIEW — NOT FOR DISTRIBUTION
Marshall researcher proposes audit-based alternative to AI proctoring for online assessments
Game-theoretic working paper suggests brief, random post-assessment conversations could deter AI-enabled cheating without surveillance software
HUNTINGTON, W.Va. — [Month Day, 2026] — As generative artificial intelligence makes it easier for students to complete online assessments without demonstrating their own learning, a new working paper from Marshall University’s David Wiley, PhD, proposes a different path: replace always-on surveillance with targeted, random learning audits.
The paper, Random Audits as a Scalable Deterrent to Cheating: Using Game Theory to Design Fair and Effective Academic Integrity Systems for the AI Era, argues that academic integrity systems should focus less on detecting every use of AI and more on changing the incentives that make cheating attractive.
Rather than relying on proctoring software, AI detectors, or lockdown browsers, Wiley proposes that instructors randomly select a small subset of students after an online exam or major project for brief learning verification conversations. Students who can explain the concepts reflected in their work keep their scores and may earn a small number of “demonstrated mastery” bonus points. Students who cannot demonstrate understanding would be referred to the course or institution’s academic integrity process.
Using a game-theoretic model, the paper suggests that a credible audit system with meaningful consequences can deter cheating at relatively low audit rates. Under one illustrative scenario — a 100-point assessment, a 25-point expected gain from AI-assisted cheating, and a 200-point penalty — the model indicates that an instructor would need to audit about 12.5% of eligible students to make cheating irrational. In a class of 30, that could mean roughly 2–4 brief conversations per assessment, depending on how the audit pool is defined.
The paper also emphasizes that fairness depends less on the audit rate than on the false positive rate: the risk that an honest student might fail an audit despite genuine understanding. To address this, Wiley recommends structured rubrics, accessibility-aware audit formats, a second-audit appeal option, and modest bonus points for students who successfully demonstrate mastery.
“The AI cheating problem is not just a technology problem; it is an incentive design problem,” Wiley said. “If students know they may need to explain their work after the assessment, and the consequences for not being able to do so are credible, cheating becomes a much less attractive strategy — without requiring invasive surveillance.”
The working paper compares the proposed system to sampling-based enforcement models used in tax auditing, financial regulation, anti-doping programs, food safety inspections, and nuclear verification. In each case, the system does not monitor every actor at all times; instead, it combines uncertainty, random checks, and meaningful consequences to sustain compliance at scale.
Wiley cautions that the proposal is not ready for broad adoption without empirical validation. The paper concludes by calling for small-scale pilot studies to test whether audit conversations are manageable for instructors, accessible and fair for students, and effective in reducing suspiciously inflated performance.
An accompanying interactive simulator allows instructors to adjust course-specific variables — including class size, audit rate, penalty level, false positive rate, bonus points, and total course points — to explore how the model behaves under different assumptions.
About the paper
Random Audits as a Scalable Deterrent to Cheating is a March 2026 working paper circulated for discussion and feedback. It proposes a sampling-based academic integrity framework for the generative AI era and recommends small pilot studies before broader implementation.
Media contact
[Marshall communications contact to insert]
Marshall University
[Email]
[Phone]
Researcher contact
David Wiley, PhD
Marshall University
Department of Marketing, MIS, and Entrepreneurship
Center for Innovation and Entrepreneurship
david.wiley@marshall.edu
https://davidwiley.org/