How Teachers Spot AI in UCAS Personal Statement Drafts

The UCAS personal statement is one of the few pieces of writing a student produces that carries real consequences beyond the classroom, and it is also one of the easiest pieces for an AI tool to draft convincingly. As Year 12 students begin their first attempts over the summer term and into the holidays, the form tutors, subject teachers, and sixth form leads who read those drafts often have to judge whether the words in front of them are genuinely the student's own. How you reach that judgement matters: the writer is usually sixteen or seventeen, a detection score is not a verdict, and the process should feel supportive rather than suspicious.

Why Personal Statements Are So Exposed to AI

A personal statement asks for a fluent, structured, reflective piece of around 4,000 characters that explains why a student wants to study a subject and what they bring to it. That is almost exactly the kind of writing large language models produce most comfortably. The genre is well documented online, the structure is predictable, and the prompts students type into an AI tool - "write me a personal statement for medicine" - return polished prose in seconds.

The pressure makes it worse. Students know the statement matters for their university offers, many feel they cannot write well about themselves, and the temptation to lean on AI for a first draft is real. A great deal of that use is not cynical cheating; it is an anxious student reaching for help with a task they find genuinely difficult. That framing matters, because it shapes how you respond when a draft looks AI-generated. The goal is not to catch a culprit. It is to make sure the statement that goes to UCAS is honestly the student's own, and to teach them why that matters before it is too late to change course.

None of this means most drafts are AI-written. The majority are not. But because the personal statement is written almost entirely at home with no supervision, running detection as a routine, transparent check protects the students who did the work themselves and gives you a fair basis for a conversation with those who may not have.

What GradeOrbit's Likelihood Score Tells You

GradeOrbit's AI detection, available on school accounts, returns a score from 0 to 100 that measures how far a draft departs from that student's own earlier writing on file. A low score means the statement sits comfortably alongside how they usually write; a high score means it differs in ways worth a closer look. The result also carries a Low, Medium or High confidence label, the signals it noticed, and a plain-English reasoning paragraph. It does not tell you with certainty that a particular student used AI, and it should never be read as a yes-or-no answer. It is an indicator to review, never proof.

For personal statements there are two things to hold in mind. First, the genre itself nudges everyone towards a similar register - reflective, slightly formal, achievement-focused - so a genuine, capable student writing earnestly about their ambitions may share surface features with AI output. A raised score on an otherwise plausible draft is a prompt to look more closely, not a conclusion. Second, a personal statement is short, which means the score is generally more useful as a flag than as proof; the longer the sample, the more informative the number. Both factors point the same way: treat the score as one input among several.

On a GradeOrbit school account, unlimited AI detection is included in every teacher seat, so there's no per-check charge and no mode to choose before running a tutor group's drafts - the same check applies whether you're screening the whole group or looking closely at one draft you intend to discuss with a student. Under your school's Data Processing Agreement (DPA), it works by comparing the draft against that student's own earlier writing on file, which is a stronger signal for a piece as consequential as a UCAS statement than a generic score would be.

Reading a High Score Fairly

A high score on a personal statement warrants a closer look, never an accusation. Your most valuable reference point is the student's own supervised writing: timed essays, in-class extended responses, reflective tasks completed in lessons. Read the flagged draft alongside that work. Does it sound like the same young person? Is the voice consistent, or does the statement suddenly acquire a polish, vocabulary, and emotional articulacy that their classroom writing never shows?

Genuine personal statements carry the texture of a real life: specific anecdotes, particular books or experiences, the slightly uneven enthusiasm of a teenager who cares about something. AI-generated statements tend to be smoothly competent but oddly generic - full of well-balanced sentences about "a lifelong passion" and "developing key skills" with very little that could only have been written by this student about this experience. A draft that is fluent but strangely impersonal, in a register the student has never previously produced, is worth gently exploring when the score is also high. As covered in our guide on how teachers handle a high AI likelihood score fairly, the number is the start of the inquiry, not the end of it.

Turning Detection Into Support, Not Accusation

If both the score and your own read raise concerns, the next step is a supportive conversation. Open with the content, not the detection result. Ask the student to talk you through a specific claim: "Tell me more about this experience you mention here - what actually happened?" or "Which part of the course are you most drawn to, and why?" A student who lived the experiences and wrote the statement can expand on them naturally. A student who generated it usually cannot add the specific, lived detail that genuine reflection produces.

Because most AI use on personal statements comes from anxiety rather than dishonesty, framing matters enormously. Make clear that the statement has to be theirs - UCAS verification and university interviews will expose a statement the student cannot back up in person - and then help them write it properly. Many students who reached for AI simply needed a way into a daunting task. Showing them how to draft from their own experiences turns a potential integrity problem into a genuine piece of teaching, and produces a far stronger statement than any AI tool would.

Keeping a Sensible Record

Personal statements do not go through the same internal-moderation cycle as coursework, but a brief contemporaneous note is still good practice, particularly where you have raised a concern with a student. A couple of sentences will do: the score, the nature of the concern, what you discussed, and the outcome. If the redrafted statement is clearly the student's own, that is the end of the matter and worth recording as such. If concerns persist, your centre's academic integrity expectations and your sixth form's reference-writing responsibilities should guide the next step, since you cannot reference a statement you do not believe is honest.

Running Drafts Through GradeOrbit

A personal statement draft is checked like any other piece of writing on a school account. Run it from the student's profile, or check a whole tutor group at once: bulk upload splits the pages by the number of pages per student and reads the name on each cover page to match it to the class list, asking you to choose when a name isn't clear. Each draft is compared with essays and written work that student has already had marked in GradeOrbit, and a student with nothing on file yet is skipped (you are told) rather than given a generic score. Typed statements go in as a PDF - export from Word or Google Docs first, as Word files aren't accepted - and handwritten plans or drafts as scans or photos, including straight from your phone via the QR camera link. Results appear on each student's profile.

How student data is handled depends on your account type: because teacher AI detection compares a draft against earlier writing on file, it runs on GradeOrbit school accounts under a signed Data Processing Agreement (DPA), where students' work is held with their real names. Uploaded scans are removed when a run archives (90 days by default; the school can choose 7 to 365 days), and student records follow the school's retention window and are wiped at offboarding. By contrast, on an individual teacher account where no school DPA is in place, uploaded student work for marking is redacted client-side in your browser, labelled anonymously, and never stored (processed and discarded immediately). Across every tier, student work is never used to train AI models.

Before the First Drafts Come In

If your sixth form plans to check drafts, say so to students at the start of the process and explain why: a statement they cannot talk about at interview does them no favours. AI detection is included in every teacher seat on GradeOrbit for Schools, with a free school trial, so a tutor team can try it on one round of drafts and see how the scores sit alongside what they know of their students.

How school accounts handle student writing, including retention, is set out at gradeorbit.co.uk/schools.