What Happens If My University Catches Me Using AI?

If your university catches you using AI, you could face anything from no misconduct finding at all to a warning, a resubmission, a mark penalty, failure of the assessment or module, suspension, or—in serious or repeated cases—exclusion. The key point is that “what happens if my university catches me using AI” is not answered by the detector score; it is answered by the assessment rules you were under, what you actually used AI to do, the evidence, and your institution’s disciplinary procedure. That distinction matters because universities are no longer treating every interaction with ChatGPT, Claude, Gemini, Copilot or Perplexity as automatically dishonest. Some courses permit AI for brainstorming, feedback, coding support, language help or research discovery. Others prohibit it for particular assessments or require a declaration. The misconduct question usually begins when AI is used beyond the permitted boundary, is not acknowledged where disclosure is required, or substitutes for work the student was meant to perform independently.

Recent evidence shows why a single dramatic answer is misleading. RTÉ reported in January 2026 that more than 500 students across responding Irish higher-education institutions had been found using AI without authorisation in graded work during 2024–25, yet sanctions varied by institution, faculty and assessment. Meanwhile, universities such as Washington State University have moved away from AI-detection software for misconduct decisions because of false-positive and fairness concerns. Turnitin’s own current guidance says its AI-writing model can misidentify human and AI text and should not be used as the sole basis for adverse action.

This guide therefore focuses on the process a student is likely to face: first, whether the AI use breached a rule; second, how suspicion becomes an investigation; third, what evidence can support or weaken an allegation; fourth, how penalties are chosen; and finally, what a student should do if questioned. It is a general guide, not legal advice, because university regulations and appeal rights differ by country and institution.

The First Question Is Not “Did You Use AI?”

The first question is whether the specific AI use was allowed for that specific assessment. A student who asks an AI tutor to explain a calculus concept and then solves the assessed problems independently is in a different position from a student who submits an AI-generated answer in an assessment that explicitly bans generative AI. The same tool can therefore be legitimate in one course and misconduct in another.

Universities increasingly describe AI use on a spectrum rather than as a simple yes/no category. Policies may distinguish brainstorming, grammar correction, translation, coding assistance, research discovery, source summarisation, drafting, rewriting and fully generated answers. Some lecturers allow several of those activities but still require disclosure. Others prohibit them because the assessment is designed to measure exactly those skills.

That is why students should treat AI as an assessment-specific tool rather than a universal study entitlement. Our guide to the best AI tools for students makes the same distinction: a tool can be useful for learning without being permitted to produce assessed work.

An allegation is therefore stronger when the university can point to a clear instruction that the student breached. It becomes more complicated when the syllabus is vague, the assignment wording conflicts with broader university policy, or the student used an adjacent technology such as grammar correction, transcription or translation that the policy does not clearly classify.

That ambiguity is not hypothetical. A 2026 study of students’ academic-integrity anxiety found that learners often received different AI rules across courses and lecturers, creating uncertainty about where acceptable assistance ended and misconduct began. A separate 2026 conceptual analysis argued that non-generative functions such as transcription should not automatically be treated as generative substitution. The practical lesson is simple: the narrower and clearer the assessment rule, the easier it is for both student and university to judge the case fairly.

Luca Iandoli, dean at St. John’s University, summarised the educational goal in April 2026: “We want to help students use AI properly, proficiently—and above all, ethically.” That framing is more useful than asking whether a university is simply “pro-AI” or “anti-AI.”

What Usually Triggers an AI Misconduct Review

A misconduct review does not always begin with an AI detector. Instructors may notice citations that do not exist, abrupt changes in vocabulary or argument quality, code the student cannot explain, factual claims that are unsupported, inconsistent formatting, impossible source access, or a final submission that differs sharply from documented drafts. A detector score can be one signal, but many institutions now treat it as only one part of a larger evidential picture.

Evidence Type	What It Shows	Typical Weight
AI detector result	A probabilistic signal that text may resemble AI-generated writing	Weak on its own; stronger only when combined with other evidence
Hallucinated citation	A reference, DOI, quotation or source that cannot be verified	Potentially significant because it is observable and assignment-specific
Version history	How the document changed over time	Can support either side by showing genuine drafting or sudden unexplained insertion
Oral discussion	Whether the student can explain arguments, sources, code or methods	Useful for testing authorship and understanding, but should be conducted fairly
Course instructions	Whether AI was prohibited, limited or required to be disclosed	Central because misconduct depends on the actual rule
Prompt/chat records	Records voluntarily supplied or required under a declared workflow	Can clarify the role AI played, subject to privacy and institutional rules

Washington State University states that suspicion is not sufficient for a finding of responsibility and says AI detection should not be the sole support for a misconduct case. In its own review of 2023–25 cases, WSU reported that 33% of review-board cases involving allegations of inappropriate AI use ended in a finding of not responsible when AI detection had been submitted without other supporting evidence.

Turnitin’s current documentation takes a similar position. It says its model may misidentify human-written, AI-generated and AI-paraphrased text and should not be the sole basis for adverse action. In August 2026, Turnitin chief product officer Annie Chechitelli also described changes to its report as intended to reduce the risk of educators overinterpreting highlighted text.

Jacqueline Evans, a psychology researcher interviewed by Nature in 2026, described a multi-signal approach: “I try to make a reasonable assessment from several pieces of information.” That is a useful description of how a defensible university review should work.

AI Detectors Can Flag Work, but They Do Not Prove Misconduct

AI-writing detectors estimate whether patterns in a text resemble material produced by language models. They do not retrieve an original ChatGPT document in the way a conventional plagiarism system can compare copied text with a known source. That technical difference is why the output is probabilistic and why false positives and false negatives matter.

Turnitin explicitly warns that its AI-writing model may be inaccurate. Washington State University cancelled its AI-detection contract in February 2026, citing false-positive risk, student distress and the number of cases in which detector-only evidence did not hold up. Penn State guidance published in 2026 similarly discouraged faculty from using detectors as determinative evidence because their assessments cannot be validated against an original source.

This does not mean a student is safe simply because detectors are imperfect. Universities can investigate through other evidence, including drafts, citations, oral questioning and the internal coherence of the work. It also does not mean every detector flag is wrong. The correct interpretation is narrower: a score is a lead to examine, not a verdict.

A 2026 AI and Ethics paper documented a broader institutional retreat from detector-centred enforcement, including universities that disabled or restricted AI-detection systems. The paper also discussed a January 2026 New York court decision annulling one university misconduct determination in a case where a Turnitin result had reported 100% AI-generated text but other evidence conflicted. That case does not create a universal rule against detectors, but it shows why procedure and corroboration matter.

The same verification problem appears in AI-assisted research. Our analysis of whether Perplexity can be wrong about cited facts explains why a citation or confidence signal is not equivalent to proof. Academic-integrity evidence needs the same discipline: trace the underlying facts rather than trusting a single automated label.

What Happens After a Lecturer Suspects AI Use

The procedure varies, but most systems contain some version of four stages: an initial concern, a request for explanation or evidence, a formal or informal decision, and a route to review or appeal. Minor cases may be handled by the instructor or department. Serious, repeated or disputed cases may be referred to a central academic-integrity office, faculty panel or conduct board.

Stage	What Commonly Happens
1. Concern is recorded	The marker identifies the assessment, suspected conduct and evidence. The student may first receive an email asking to discuss the work.
2. Student is asked to respond	The student may be invited to explain the writing or problem-solving process, provide drafts, discuss sources, or answer questions about the submitted work.
3. Policy is applied	The decision-maker considers the exact assessment instructions and the university’s academic-integrity rules, not merely whether AI was involved.
4. Finding and sanction	If misconduct is established, the decision normally states the breach and penalty. If it is not established, the allegation should not be treated as proven.
5. Review or appeal	Most universities provide a procedure to challenge factual error, procedural unfairness, disproportionate penalty or other specified grounds.

Some institutions are deliberately making first-time minor cases more educational. Cornell University announced in August 2026 that its Accepting Responsibility programme would expand across the university after a pilot, allowing some minor first-time violations—including certain prohibited AI uses—to be resolved without a full primary hearing. The programme combines responsibility with education rather than treating every case as the most serious form of cheating.

Liz Karns, Cornell’s director of academic integrity initiatives, explained one practical reason: “Primary hearings take a substantial amount of time.” The wider point is that universities often distinguish lower-level cases from serious or contested allegations rather than routing everything through the same process.

At the other end of the spectrum, a case involving a prohibited exam, substantial outsourcing of intellectual work, repeated misconduct, fabricated evidence, impersonation, contract cheating or deception during the investigation can move quickly into a more serious disciplinary track.

Possible Penalties: From Resubmission to Exclusion

There is no universal penalty table for AI misuse. A university’s regulations may give departments discretion within a range, and the sanction may depend on the assessment’s value, the amount of unauthorised assistance, intent, prior findings and whether the student admitted or contested the conduct.

Possible Outcome	When It May Arise	What It Can Mean
No misconduct finding	Evidence is insufficient, use was permitted, or the allegation is not established	No academic penalty for the allegation
Educational warning or required training	Low-level or first-time issue under some systems	Integrity module, meeting, written warning
Resubmission / capped resit	Work cannot be accepted as submitted but remediation is permitted	Rewrite or resit, sometimes with a mark cap
Mark reduction / zero on assessment	Unauthorised AI materially affected a piece of assessed work	Reduced mark or zero for that task
Module/course failure	Serious breach or policy assigns a fail-level penalty	Fail grade for module/course
Suspension	Serious or repeated misconduct	Temporary removal from studies
Exclusion / expulsion	Most serious or repeated cases under institutional rules	Termination of enrolment

RTÉ’s January 2026 survey illustrates that variation. It reported that some students received an F on affected coursework with resit options, while other institutions used grade penalties, assignment failure, capped resubmissions or, for repeated unauthorised AI use, the possibility of suspension. Because many institutions did not separately record AI cases, the available numbers were incomplete.

An international example shows how severe the top end can be. Research describing Chinese University of Hong Kong guidance noted that serious academic dishonesty penalties may include assessment review, permanent demerits, course failure, suspension, lower degree classification, expulsion or even revocation of academic certification under applicable procedures. That does not mean such outcomes are typical for a first AI case; it shows the ceiling that some academic-integrity regimes reserve for serious dishonesty.

The practical mistake is assuming either extreme: “I will definitely be expelled” or “they can only give me a warning.” Neither is reliable without the institution’s regulations. Read the penalty provisions and any published precedents or student-handbook guidance for your own university.

What Makes a Case More or Less Serious

Decision-makers commonly look beyond the fact that an AI tool was used. The seriousness of the breach usually turns on what the tool replaced, how much of the assessment was affected and whether the student tried to conceal or misrepresent the process.

Category	Examples
Lower seriousness factors	AI used for a narrow permitted-adjacent task; unclear guidance; first incident; small part of assessment; prompt disclosure; genuine drafts; cooperative explanation
Higher seriousness factors	AI generated core assessed content; explicit prohibition; exam use; fabricated citations or evidence; repeated misconduct; deception; inability to explain submitted work
Context that may matter	Disability adjustments, language-support rules, conflicting instructions, tool functionality changing mid-course, supervisor or lecturer guidance

Thea Goldring, a lecturer in Princeton University’s Writing Program, told Nature that in her course an authorship discussion can affect the grade when students cannot demonstrate mastery: “if students can’t demonstrate authorship and mastery during the discussion with me, it caps their grade at a C.” Her example is course-specific, not a universal university rule, but it captures a broader principle: assessors may care about whether the submission represents the student’s own understanding.

That principle also explains why fabricated references are especially damaging. An AI-generated citation that does not exist is not simply a style problem; it can show that the student submitted claims without checking their provenance. A student who used AI for source discovery but independently verified every reference is in a materially different evidential position.

For research-heavy assignments, our AI research assistant comparison stresses source traceability and verification because AI-generated bibliographies can look convincing even when individual references are wrong.

If You Are Accused, Preserve Evidence Before You Panic

If you receive an allegation, the most useful immediate step is to preserve the real record of how you worked. Do not rewrite your document history, delete chats that may be relevant, manufacture drafts or ask an AI system to create a false explanation. Those actions can turn a debatable authorship question into a credibility problem.

• Save the version history of the document, including cloud revision records where available.
• Keep outlines, reading notes, source PDFs, citation-manager records, calculations, code commits and earlier drafts.
• Keep any AI conversation that genuinely shows a permitted or limited use, if disclosure is lawful and relevant.
• Download the assessment brief, syllabus and AI policy exactly as they existed when you submitted.
• Write a factual timeline of your process while details are fresh, separating your own work from any AI assistance.
• Read the allegation carefully and identify what conduct is actually alleged, rather than responding to a broader fear that “AI was detected.”

If you did use AI beyond the permitted boundary, a truthful response is usually safer than an invented technical defence. You can still explain context: what you used, how much, what you understood the rule to mean, whether you verified the output, and whether the final work genuinely reflects your knowledge. The university will decide what weight those facts carry.

If you did not use prohibited AI, the goal is not to “beat” the detector. It is to show authentic process. Draft history, contemporaneous notes, browser or library research records, citation-manager entries, data files, handwritten work, coding history and the ability to explain decisions can be more persuasive than running the essay through several competing detectors.

Student-union advisers, academic advisers, ombuds services or independent student advocates can also help interpret procedure. Where the outcome could affect immigration status, professional registration, a funded placement or a degree award, consider obtaining advice suited to your jurisdiction.

What Not to Do During an Investigation

An accusation can create pressure to act quickly, but several responses can make the situation worse. The safest approach is to preserve the original record and respond to the actual allegation.

• Do not fabricate drafts, timestamps, citations, prompts or screenshots.
• Do not delete relevant evidence after being notified of an investigation if university rules require preservation.
• Do not submit another AI-generated statement pretending to be a personal account of your process.
• Do not claim that AI detection is always invalid; the stronger argument is that detector evidence must be interpreted with corroboration and policy context.
• Do not assume that a first meeting is necessarily a final hearing; check the procedure and deadlines.
• Do not miss appeal or review deadlines while trying to negotiate informally.

Also be cautious with privacy. Uploading a university’s confidential allegation letter, another student’s work, unpublished research data or personal information into a consumer chatbot may create a second problem. If you use AI to help understand a policy, redact names, student numbers, private feedback and sensitive material unless your institution expressly permits that workflow.

Our guide on whether it is safe to paste personal information into AI tools explains why “not used for training” and “not stored” are different privacy promises.

How Universities Are Moving Beyond Detector-Only Enforcement

The 2026 direction of travel is not that universities have stopped caring about AI misuse. It is that many are looking for evidence of learning, process and authorship that is more defensible than a single classifier score. This includes oral follow-ups, in-class work, staged drafts, source annotations, process reflections, version histories and assessment designs that make the student’s reasoning visible.

Cornell researchers reported in May 2026 that an analysis of more than 95,000 students at 20 U.S. public research universities found roughly one-third regularly used generative AI for assignments and 9% reported using it to cheat. Study co-author Rene Kizilcec’s conclusion was concise: “Assessment reform is necessary and urgent.” The finding matters because enforcement alone cannot resolve a situation in which legitimate and illegitimate AI use are both widespread.

Brown University’s 2026 teaching guidance similarly framed the challenge around protecting learning and critical thinking while adapting assessment. Instructors interviewed by Nature described using oral explanation, in-person work and multiple evidence points rather than assuming detectors can settle authorship.

For students, this shift cuts both ways. It can reduce the risk of being judged solely by an unreliable score, but it also makes superficial AI outsourcing easier to expose. If you cannot explain the logic, sources, code, calculations or wording choices in your own submission, an oral check can reveal that gap even when no detector is used.

This is also why source-grounded tools do not remove student responsibility. Our guide to the best AI tools for reading research papers treats AI summaries as a reading aid, not a substitute for checking the paper’s actual methods and claims.

A Practical Boundary: Study Support vs. Substitution

A useful way to judge risk before submission is to ask what cognitive work the assessment is trying to measure. If the task tests your ability to form an argument, write code, interpret data, translate a passage, critique a paper or produce original prose, using AI to perform that exact function is more likely to cross the line when permission is absent. Using AI to explain a concept before you independently perform the assessed task may be compatible with learning, depending on the rules.

AI Use	Typical Risk	Why It Matters
Explain a concept in simpler terms	Often lower risk for private study	Still verify accuracy and follow course rules
Brainstorm possible essay questions	Often lower to moderate	Do not submit generated ideas as original if disclosure is required
Suggest grammar improvements	Policy-dependent	Some courses permit; others treat rewriting tools as assistance requiring disclosure
Generate paragraphs for submission	High risk when unauthorised	May substitute for assessed authorship
Generate code for a programming assessment	High risk when independent coding is being assessed	Permission varies sharply by course
Create references or quotations	High academic risk	AI can fabricate sources; every citation must be verified
Answer a closed-book or no-AI exam	Very high risk	Usually directly conflicts with assessment conditions

This is not a universal policy table; it is a reasoning tool. Your university’s instructions control. A lecturer may explicitly require students to use AI and critique its output, which reverses the normal risk. Another may ban all generative assistance because the assessment is designed to test unaided performance.

For evidence-heavy coursework, the safest workflow is to use AI for discovery and explanation while preserving a traceable research chain. Our best AI tools for research guide explains why no single assistant can replace primary-source verification.

Can a University Punish You If the Policy Was Unclear?

A vague or changing AI policy does not automatically invalidate an allegation, but clarity matters. Fair disciplinary systems should identify the rule, explain the evidence, give the student an opportunity to respond and apply the published procedure. If course instructions were genuinely contradictory, that can be relevant to both responsibility and sanction.

Students should distinguish three arguments. First, factual denial: “I did not do what is alleged.” Second, rule interpretation: “I used the tool, but the instructions permitted or reasonably appeared to permit that use.” Third, mitigation: “I breached the rule, but there are circumstances relevant to the penalty.” Mixing all three into one vague defence can weaken a response.

The strongest cases are document-based. Save the syllabus version, learning-platform announcement, assignment brief, AI declaration instructions and any email or forum answer from the lecturer. If a teacher told the class that grammar assistance was allowed, the exact wording matters more than a later general statement that AI is discouraged.

Procedural fairness also matters when universities use automated tools. The 2026 research literature on detector bias and false positives has increased pressure on institutions to avoid treating opaque probability scores as self-explanatory evidence, particularly for multilingual writers. That does not guarantee a student will win an appeal, but it makes the quality of the university’s reasoning and corroboration important.

How to Use AI Without Creating an Academic-Integrity Problem

The safest strategy is not to search for a detector-proof writing method. It is to build an auditable learning process that stays inside the permission you have been given.

• Check the assessment brief first, not after you finish.
• If the rule is unclear, ask the lecturer a narrow question such as whether brainstorming, grammar correction or code completion is permitted.
• Keep your own outline, notes and source trail so you can demonstrate authorship.
• Verify every factual claim, quotation, DOI and reference against the original source.
• Use AI-generated explanations to support learning, then close the tool and perform the assessed reasoning yourself when required.
• Disclose AI use in the format your institution specifies, including tool, purpose or prompts where required.
• Avoid uploading confidential research, personal data, exam material or copyrighted course resources into external AI systems without permission.

These habits are not merely defensive. They improve the quality of the work. AI systems can produce fluent but unsupported text, generate citations that do not exist, merge details from different sources and overstate uncertain conclusions. A student who keeps the original evidence chain is less likely to submit those errors.

That is also the best preparation for an oral authorship check. If you can explain why you chose a source, how your argument changed, where a calculation came from and what the limitations are, your process is visible. If an AI tool did most of that reasoning and you only edited the surface language, the gap is much harder to hide.

Our Editorial Verification Process

This explainer was researched in September 2026 using current university policies, official detection-tool guidance, 2026 higher-education reporting and peer-reviewed research. We checked Washington State University’s current AI-misconduct guidance, Turnitin’s AI Writing Report documentation and August 2026 product update, Cornell University’s 2026 academic-integrity research and Accepting Responsibility programme, Brown University’s August 2026 teaching guidance, RTÉ’s January 2026 survey of Irish institutions, and 2026 peer-reviewed work on student anxiety, detector limitations and academic-integrity policy.

For the search landscape, we reviewed exact and near-exact ranking pages for the query “what happens if my university catches me using ai.” The dominant structures focused on punishment lists, detector explanations and short “what to do if accused” checklists. We deliberately used a different architecture built around the decision chain: permission, evidence, procedure, severity, response and future-safe use. This makes the article more useful to a student who needs to understand not only possible sanctions but how a finding is actually reached.

The Perplexity AI Magazine sitemap endpoints specified in the editorial brief did not return parseable XML through the available browsing layer. Rather than inventing a sitemap inventory, the internal links in this article were selected from live indexed Perplexity AI Magazine pages with direct relevance to student AI use, academic research, source verification and AI privacy.

We do not claim that any detector score can establish authorship, nor that the penalty examples in this article predict what a particular university will do. Institution-specific regulations remain the controlling source. Turnitin institutional pricing is not publicly standardised on the reviewed official pages, so no commercial price was inferred.

This article was researched and drafted with AI assistance and reviewed by the Sami Ullah Khan editorial desk at Perplexity AI Magazine. All data, citations, pricing figures, and named quotes have been independently verified against primary sources before publication.

Conclusion

If your university catches or suspects you using AI, the outcome is not determined by the word “AI” and it is not determined by a detector percentage. The decisive questions are what the assessment allowed, what the tool actually did, what evidence supports the allegation, whether the process was fair and how serious the breach is under the institution’s rules.

That produces a wide range of possible outcomes. A permitted use may result in no case at all. A minor first-time breach may lead to education, resubmission or a mark penalty. More serious unauthorised substitution can lead to assessment or module failure, while repeated or severe misconduct can reach suspension or exclusion. The strongest 2026 guidance increasingly treats automated detection as one clue among several rather than a standalone verdict.

For students, the durable protection is process. Read the exact assessment instructions, preserve your drafts and sources, verify AI output, disclose assistance when required and make sure the submitted work is something you can explain. Universities are still rewriting their AI rules, so open questions remain about consistency, detector use and the boundary between ordinary digital assistance and generative substitution. But the direction is clearer: academic integrity is moving toward demonstrable authorship and learning, not simply a machine-generated score.

FAQs

Q: What happens if my university catches me using AI?

A: You may face no misconduct finding, a warning, resubmission, mark penalty, failure of an assessment or module, suspension, or—if the case is serious or repeated—exclusion. The result depends on the assessment rules, what AI did, the evidence, your previous record and the university’s disciplinary procedure.

Q: Can a university fail me for using ChatGPT?

A: Yes, if your use breached the assessment or academic-integrity rules, a university may impose a fail-level penalty on the assignment or module. That is not automatic. The institution should apply its published procedure and penalty framework to the facts of the case.

Q: Can Turnitin prove I used AI?

A: No. Turnitin says its AI-writing model may misidentify human-written and AI-generated text and should not be used as the sole basis for adverse action. A university may still use the report as one signal alongside drafts, citations, version history, oral explanation and other evidence.

Q: What if I used AI only for grammar?

A: It depends on the course. Some assessments permit grammar or language support; others require disclosure or restrict rewriting tools because language itself is being assessed. Check the exact instructions rather than assuming grammar assistance is automatically allowed.

Q: What should I do if I am falsely accused of using AI?

A: Preserve drafts, version history, notes, source records and other evidence of your process. Read the allegation and policy, respond factually, and use student-advice or appeal channels where available. Focus on authorship evidence rather than trying to produce a competing detector score.

Q: Can I be expelled for a first AI offence?

A: Expulsion is possible under some university misconduct frameworks but is generally at the severe end of sanctions. A first, limited case may be treated less harshly than repeated misconduct, prohibited exam use, substantial outsourcing, fabricated evidence or deception.

Q: Do I have to show my ChatGPT history to my university?

A: That depends on your institution’s rules, the investigation process and local privacy law. Do not assume you must surrender an entire private account. Read the formal request carefully and seek student or legal advice if it is broad, sensitive or unclear.

Q: Is using AI for research academic misconduct?

A: Not necessarily. Many institutions allow AI for research discovery, explanation or planning, but require students to verify sources and follow disclosure rules. It becomes risky when the tool performs prohibited assessed work, generates unverified citations, or is used contrary to explicit instructions.

References

• Alalwani, S. (2026). Faculty perceptions of ChatGPT on academic integrity and institutional roles in higher education. Scientific Reports, 16, 27532.
• Chechitelli, A. (2026, August 4). How Turnitin is simplifying AI detection for educators and publishers. Turnitin.
• Cornell University. (2026, May 21). Widespread AI misuse means higher ed must rethink assessment.
• Cornell University. (2026, August 12). New program turns minor cheating violations into a learning opportunity.
• Nature. (2026). From Trojan horses to AI-proof exams: How professors are tackling students’ AI use. Nature, 658, 279–280.
• RTÉ News. (2026, January 11). Over 500 college students found using AI illegally in coursework.
• Turnitin. (2026). Using the AI Writing Report. Turnitin Guides.
• Washington State University. (2026). Detecting and reporting misconduct related to generative AI. Office of the Provost & Executive Vice President.
• Wright, C. (2026). Transcription is not generation: Distinguishing non-generative AI tool use from academic misconduct in higher education assessment. International Journal for Educational Integrity, 22, 24.