How Teachers Spot ChatGPT in Student Essays: A UK Classroom Guide
By Tomas Lerpiniere · Updated 2026-10-28
Most teachers recognise the moment: a Year 10 or Year 13 draft suddenly sounds like a textbook abstract. This guide lists practical linguistic signs, explains why EAL and scaffolded writers get false positives, and gives a four-step check-in script that keeps integrity conversations fair.
In this guide
Feedback Flows tools in this guide
Linguistic tells of LLM prose vs authentic teen writing
- Uniform burstiness. Sentences of near-identical length and cadence across a paragraph.
- Stacked transitions. Furthermore / Moreover / In essence / Holistic approach in consecutive sentences.
- Generic historiography or criticism. Sweeping claims with no named lesson, text extract or classroom task.
- Sudden register shift. A learner who writes short, concrete sentences overnight produces dense nominalisations.
- Bibliography red flags. Plausible-looking titles with wrong years or DOIs that do not resolve — see how to spot fake AI-generated references.
None of these alone proves misconduct. Paste the section into the free AI writing detector to see which sentences the hybrid engine highlights, then compare coverage with the overall score.
EAL writers and PEEL/PETAL frames
EAL learners and classes taught heavily scaffolded PEEL/PETAL paragraphs often reuse safe connectives and even pacing. Statistical detectors can misread that as AI. Before escalating:
- Select EAL or Younger writer context on the detector.
- Ask whether the paragraph matches a frame you taught that week.
- Invite an oral paraphrase — authentic writers can usually unpack their own points.
For JCQ authentication stages and centre reporting, read JCQ AI guidance for NEA and coursework. For why lone percentages are weak evidence, see evidence over AI detection scores.
A four-step non-accusatory check-in
- Open with curiosity. “I’d like to talk through this paragraph so I understand how you built it.”
- Show the quotes. Display highlighted sentences, not a raw percentage.
- Ask process questions. Where did the idea start? Which source supports this claim? Can you explain this term in your own words?
- Agree next steps. Support and redraft, further authorship/citation checks, or escalation under policy — then record the outcome.
When voice diverges from earlier authentic work, add writing fingerprints or the free authorship checker. Full workflow detail lives in writing fingerprints and authorship checks in UK schools.
Free tools to gather evidence quickly
- Free AI writing detector — sentence highlights and writer context
- Free authorship checker — style consistency vs a baseline sample
- Free citation checker — Crossref/OpenAlex bibliography screening
- Free collusion checker — side-by-side overlap between two submissions
Department-wide expectations belong in your school AI policy, not in ad-hoc hallway conversations.
FAQ
Related guides
- Evidence Over AI Detection Scores Why a single AI percentage is a weak basis for a conversation, and what better integrity evidence looks like.
- Writing Fingerprints and Authorship Checks in UK Schools How UK schools, colleges and ITPs can investigate suspected AI work fairly with writing fingerprints, process evidence and JCQ-aligned workflow.
- JCQ AI Guidance for NEA and Coursework JCQ AI use in NEA and coursework for UK schools: authentication stages, acceptable AI, and evidence for fair centre investigations.
Continue exploring