How Teachers Detect AI in GCSE English Coursework

GCSE English Language and Literature coursework has always attracted a degree of concern around academic integrity. But the arrival of tools like ChatGPT and Claude has changed the nature of that concern entirely. Where teachers once worried about students copying from SparkNotes or commissioning essays from essay mills, the challenge now is more subtle: a student who uses an AI writing tool can produce something that reads as original, uses their own prompt, and is technically in their own words - yet is not genuinely their own work.

Since the 2015 reforms, GCSE English Language has no written coursework: its only non-examined assessment is the Spoken Language endorsement, which is reported separately from the grade. The written work English teachers set outside the exam - practice creative and transactional writing, homework essays, mock responses - still exists to show a student's own voice and development, and AI can imitate that convincingly. That is why a detection result has to be read alongside what you know of the student.

Why GCSE English Coursework Is a High-Risk Area

English coursework is particularly vulnerable to AI use for several reasons. First, the tasks are well-defined and publicly available - AQA and Edexcel past papers and specimen materials are widely accessible online, and AI tools may well have been trained on them. A student can prompt an AI with the exact task wording and receive a highly polished response.

Second, GCSE English writing tasks reward exactly the qualities that AI models are strong at: coherent structure, controlled vocabulary, varied sentence lengths, and topic development. A Year 11 student who struggles with writing but can use a phone effectively can produce work that reads like that of a confident, capable writer.

Third, unlike a handwritten exam answer, coursework submitted digitally leaves no physical trace of the drafting process. Without draft evidence, classroom observations, or teacher knowledge of the student's written voice, it can be extremely difficult to identify AI use through reading alone. That is where AI detection tools become useful.

How Likelihood Scores Work

AI detection tools do not read student work the way a teacher does. They analyse statistical patterns - the predictability of word sequences, syntactic regularity, vocabulary distribution, and structural uniformity - and compare them against characteristics common in AI-generated text. The result is a likelihood score: a number from 0 to 100% that represents how closely the text resembles AI-generated content.

A score of 80% does not mean a student definitely used AI. It means the text shares significant statistical features with AI output. High scores on AI-generated writing exist on a spectrum, and the score alone should never be the basis for a formal action. False positives do occur, particularly for students who are strong, precise writers - or who have followed modelling so closely that their style has become algorithmic-looking.

GradeOrbit's AI detection works differently: rather than comparing a piece with AI writing in general, it compares it with that student's own earlier work on file, and returns a score for how far the new piece departs from it alongside a breakdown of the specific signals identified. This transparency is important: it gives teachers something concrete to discuss with a student and with line managers, rather than an opaque verdict.

Reading the Signals Beyond the Score

When you receive a high likelihood score on a piece of GCSE English coursework, the most useful thing you can do is consider it alongside everything else you know about that student.

• Is this consistent with the student's performance in timed, in-class writing tasks?
• Does the voice in the submission match the voice you hear from this student in lessons?
• Is the vocabulary and syntax notably above the student's usual standard?
• Did the student produce a draft? Is there a record of the drafting process?
• Has the student discussed their ideas verbally with you or in peer review?

In many cases, a high score will prompt you to look at a piece of work more carefully - and you will find that contextual evidence either supports or contradicts the concern. A high score on work from a student who consistently performs at that level, produces drafts, and engages thoughtfully in class discussions should be treated very differently from a high score on work from a student who rarely hands in written tasks and submitted this one late.

If you are uncertain how to interpret the result of a detection check, the guide on how to interpret AI detection scores covers the decision-making process in detail.

Using GradeOrbit's Detection Tool for English NEA

AI detection is part of GradeOrbit's school accounts. Run it from a pupil's profile for one piece, or on a whole class at once. Upload scans, photos or PDFs, or use the phone QR camera link; for a class set, bulk upload splits the pages by "pages per student" and matches the name on each cover page to your class list, asking you to pick the pupil if the match isn't clear. Typed NEA goes in as a PDF or image (export Word documents to PDF first). The tool returns a score, a confidence label, the signals it noticed and a plain-English reasoning paragraph on the pupil's profile.

AI detection is included with unlimited checks in GradeOrbit's School teacher seats, so there is no separate tier to weigh up between a quick scan and a more thorough pass and no per-check charge - departments can afford to check every submission that raises a flag rather than rationing checks. On School accounts operating under a signed Data Processing Agreement (DPA), coursework is compared against that pupil's earlier work on file within your school's agreed retention window, giving you a grounded comparison rather than an isolated score; a pupil with no earlier work on file is skipped, and you are told. Individual teacher accounts focus on marking and do not include AI detection, and submissions across all tiers are never used to train AI models.

For GCSE English coursework, it is worth running a check on any submission that raises a flag, particularly pieces you are using as evidence of progress or to inform predicted grades. Documentation is important: save the score and breakdown alongside your professional notes before taking any action.

What to Do When a Score Is High

A high likelihood score is the start of a process, not the end of one. The appropriate response is a professional conversation, not an immediate referral. Ask the student to talk you through their work: what they were trying to do, where their ideas came from, what they found difficult. Ask them to complete a short related writing task in a supervised setting. In most cases, this will give you the clarity you need.

Your school's academic integrity policy and the JCQ guidance on AI use in assessments should define what happens next. AQA and Edexcel both have clear procedures for suspected malpractice that your exams officer will be familiar with. Never act on a likelihood score alone - always document your reasoning and ensure that any formal process involves the student being given the opportunity to respond.

It is also worth taking a proactive approach at the start of NEA season. A clear, plain-English explanation of what constitutes AI misuse - delivered to students before they begin their coursework - reduces the likelihood of dishonest submissions and protects students who might otherwise stumble into a policy breach without fully understanding the rules.

Staying Consistent Across Your Department

One of the risks of introducing any new tool is inconsistency. If one teacher checks for AI use and another does not, or if thresholds for action vary across the department, students are not being treated equitably. This matters particularly for a qualification component like the GCSE English NEA, where the outcome has long-term consequences.

Agree as a department on when detection checks will be run, what score threshold triggers further action, and how results will be recorded. Make sure all staff understand that the score is advisory and that professional judgment is always required. A moderation meeting at the start of NEA season is a good opportunity to establish shared norms and review any early cases together.

Before NEA Season Starts

Agree the threshold, the record-keeping and the conversation process as a department, explain the rules to students before they begin, and treat every score as a reason to look more closely rather than a finding. That approach protects students as much as it protects the qualification.

For departments considering it, GradeOrbit for Schools includes unlimited AI detection alongside essay marking, and you can start with a free school trial.