A George Washington University student accused of AI use after a paper triggered a detector flag faces the university's Code of Academic Integrity process, not a summary judgment. Understanding what the code actually requires, and what the research says about detector reliability, changes the shape of a response. The defense package begins with those two facts.
The pattern at George Washington University
GW AI detection accusations follow a shape now well documented across institutions. An instructor runs a submitted paper through Turnitin's AI writing indicator or a similar tool. The score exceeds whatever threshold the instructor treats as significant. The instructor either raises the concern directly with the student or refers the matter to the university's academic integrity process under the Code of Academic Integrity.
At GW, allegations are handled through the Office of Student Rights and Responsibilities. The code applies to all schools at the university, and the process moves from an initial notice through an informal resolution meeting or a formal hearing before an Academic Integrity Panel. A detector score is often the entire evidentiary basis for the initial referral, which is a structural problem this post addresses directly.
Why detectors flag human writing
AI detectors measure statistical properties of text, mainly perplexity (how predictable the next word is) and burstiness (variation in sentence length). Writing that is careful, formal, or structurally consistent tends to score as more "AI-like" by these metrics, regardless of who produced it.
Peer-reviewed research has consistently found meaningful false positive rates. Weber-Wulff et al. (2023), published in the International Journal of Educational Integrity, tested fourteen detection tools and concluded that none met a reliability standard suitable for high-stakes decisions. Liang et al. (2023), in Cell Press Patterns, found detectors flagged writing by non-native English speakers at sharply higher rates than writing by native speakers. A deeper walkthrough of that research is in the research summary on detector false positive rates.
Turnitin itself has publicly acknowledged elevated false positive rates on documents with lower AI-flagged percentages and has told institutions the AI indicator score should not be used as sole evidence of misconduct.
What the GW Code of Academic Integrity requires
The Code of Academic Integrity at George Washington University defines academic dishonesty to include "cheating," which encompasses unauthorized use of external assistance on academic work. Whether AI use counts as prohibited assistance depends on the specific course, the assignment, and the instructor's stated policy. There is no blanket, university-wide rule that every AI use is a violation.
Procedurally, the code entitles the accused student to written notice of the specific allegation, access to the evidence the university intends to rely on, an opportunity to respond, and either an informal resolution or a formal hearing before an Academic Integrity Panel. Findings must be supported by a preponderance of the evidence: it is more likely than not that a violation occurred. A detector score alone, without corroborating evidence, is a thin foundation for that standard.
Questions to ask in writing
- Which detection tool produced the flag, and what was the specific score?
- What threshold does the instructor or the university treat as evidence of AI use, and where is that threshold documented?
- Did a human review the flagged passages before the referral was made?
- What non-detector evidence supports the allegation?
- What is the course's written policy on AI tool use, and where does the syllabus define the boundary?
Evidence that shifts these cases
Process evidence is what closes AI detection cases. That means artifacts showing how the paper was actually written, over what period, and with what sources.
| Evidence | What it shows |
|---|---|
| Google Docs or Word version history | The paper was drafted incrementally, over time, with revisions consistent with a human writing process |
| Browser history from the drafting period | Which sources were consulted, when, and for how long |
| Handwritten notes or outlines (photographed with timestamps) | Pre-drafting thinking that maps onto the final structure of the paper |
| Library database access records | Genuine engagement with the sources actually cited |
| Comparison scores from other detectors | Inconsistency across tools, which undermines the reliability of any single score |
If this is you at George Washington University
Move deliberately. The first written response to the instructor or to the Office of Student Rights and Responsibilities frames the record for everything that follows.
- Read the notice of allegation carefully and confirm the specific charge and the process track (informal resolution or formal panel).
- Request, in writing, the detector name, the specific score, and any non-detector evidence supporting the allegation.
- Preserve drafts, version history, browser history, and any handwritten notes before retention windows close.
- Read the course syllabus alongside the Code of Academic Integrity to confirm what was actually prohibited.
- If the proposed sanction is suspension, expulsion, or a transcript notation, or if you are an international student whose visa status depends on enrollment, consult an education law attorney before any hearing.
If you are preparing a written response, NotBot generates a personalized defense package that names the detector, cites the research on false positives, and documents your specific writing process. If a finding has already been issued, the appeal package is built around the grounds most GW panels actually consider.
Build your GW defense package
A personalized response that documents your writing process and addresses the detector that flagged your work.
Get your defense package$49 one-time · Generated in 60 seconds