University Asked About AI in Your SOP? Next Steps
A university emailed about AI in your statement of purpose. What its rules let it do, the evidence to gather today, and a reply you can adapt.
University Asked About AI in Your SOP? Next Steps
Reply by the deadline in the email, in plain facts, and offer to discuss your statement live. Do not build your reply around detectors, and do not send a rewritten statement. Under the rules quoted below, a university can cancel an application or withdraw an offer, and it decides on what you show it.
Our view: if you wrote your statement of purpose yourself, your strongest answer is a conversation, not an email. The best evidence is that you can explain and defend every claim in it, live. UBC says programs may ask you to express your ideas aloud, and Cambridge and UWE say interviewees confirm they are not using AI. Dated drafts help; your ability to talk through paragraph three helps more.
Every university page quoted below was checked on October 31, 2026. None of them states a cycle year, so reread your own program's rule and the declaration you signed in its portal before you reply. Oxford's graduate AI page returned a bot check (HTTP 403) that day and is not quoted.
Read the email for what it actually asks
We found no published template for these emails, so sort yours by what it asks you to do.
| The email asks you to… | What it is | Your move |
|---|---|---|
| Explain how you wrote the statement, or confirm it is your own | A request for information | Reply in writing with facts and a short timeline; offer a call |
| Attend an interview, a call or a timed writing task | A test of your "written and verbal expression" | Accept the first slot you can; prepare as below |
| Tell you of a decision (application canceled, offer withdrawn) | A decision | Ask, once and politely, whether there is a review process and what it requires |
What the published rules let a university do
UBC's applicant declaration has the fullest list of consequences among the pages we checked. Each UBC applicant submits it:
I agree that if I knowingly or carelessly provided untrue, incomplete, fraudulent or plagiarized information with this application then UBC may in its sole discretion do any or all of the following: (a) cancel my application; (b) withdraw any offer of admission, whether accepted or not; (c) require me to withdraw from UBC; (d) subject me to academic discipline; (e) share the information I provided with other post-secondary institutions, law enforcement agencies, or other third parties.
Other universities are shorter. The University of Alberta: "Any detected use of AI-generated content will be considered a violation of academic integrity and may result in application rejection or other consequences." TUM: "The use of AI tools such as ChatGPT to create documents for an application at TUM is not permitted. Please refrain from using AI when creating your documents. Self-created application documents are thoroughly checked for the use of AI. Your personal suitability for admission will be checked during the application process. If the use of AI applications for the creation of your application documents is detected, you will be excluded from the application process, i.e. your application will be rejected for deception." Cambridge treats AI-invented achievements as fraud, and its fraud page says "we will withdraw any other applications you have made to the University."
These rules differ by university and by country. AI rules for UK master's personal statements has them side by side, and do grad schools check SOPs for AI? covers US attestations.
Why a flag is not proof
A query can start from a reader's doubt or a software signal. Neither is a finding until the university decides.
If you are tempted to quote a detector result in your reply, read SOP AI detector percentage: is any number safe? first.
UBC's graduate guidance, in a section on detecting student use, says: "The use of applications to detect AI-generated content is strongly discouraged at this time, due to concerns about effectiveness, accuracy, bias, privacy, and intellectual property." A 2023 study by Liang and colleagues found that the AI-writing detectors it tested "consistently misclassify non-native English writing samples as AI-generated, whereas native writing samples are accurately identified" (arXiv). If the email mentions software and English is your second language, one sentence pointing to that study, with the link, is enough. Then move on to your evidence. A university is not obliged to accept the study as an answer.
What none of the pages above says is that a score alone decides. The rules talk about "detected use", "untrue" information and "deception". Your job is to show the work behind the statement.
Evidence to gather today
Collect it before you reply, while the dates are easy to find.
- Version history. Your editor's file history, with dates. Export or screenshot the earliest version and two or three later ones.
- Notes and outlines. Handwritten notes, phone notes, a list of facts you started from.
- Emails. Messages to recommenders, friends or advisers that mention or attach drafts.
- The record behind each claim. The project report, code repository commits, the transcript line, the internship letter, the paper. If paragraph two says you cut processing time by a third, find the file that shows it.
- What you read about the program. The course page or lab page your fit paragraph draws on.
What does not help: re-running the statement through a checker, rewriting it so it "passes", or sending a new version. A new version answers a different question from the one you were asked.
Constructed example: a three-line evidence timeline
Constructed example, written by us, not a real applicant's record:
- August 4, 2026: first outline in notes app (screenshot attached), listing the traffic-sensor project and the CGPA explanation.
- September 12, 2026: full draft, 1,140 words (file history screenshot attached).
- October 20, 2026: final draft, 980 words, after comments from Dr. S. Menon (email attached).
A reply you can adapt
Constructed example, written by us, not a real university email or applicant reply.
The query (invented, 43 words):
Dear Applicant, as part of our review we would like to understand how your statement of purpose was prepared. Please describe the process you followed, including any tools used, within five working days. We may also invite you to a short online interview.
The reply (invented, 113 words):
Dear Admissions Committee,
Thank you for your message. I wrote my statement of purpose myself, from my own notes, between August and October 2026. A short timeline with screenshots of my drafts is attached.
The only tool I used was the spell checker in my word processor. I did not use AI to generate, rewrite or translate any part of the statement.
The statement's main claims come from my final-year project on traffic-sensor data and my internship at a logistics firm. I have attached the project report and my supervisor's email about the draft.
I would welcome an interview or a supervised writing task if that would help.
Kind regards, Ananya Rao
Why it works: it answers the question asked, lists tools honestly, points each claim to a document, and offers the live test the rules describe. It does not argue about software. Change every detail to your own; if any sentence in it is not true for you, delete it.
If you did use AI
Do not deny it. Say what you used, for which part, and how that fits the program's rule. A false answer is a second problem on top of the first: UBC's declaration covers "untrue" information as well as plagiarized information.
Check your use against the rule you signed. UBC says: "Any text produced by GenAI that appears in the application package must be cited, and the process used made transparent." Cambridge says: "AI must not be used when writing personal statements and CVs. This prohibition covers the use of AI to help your use of English." If the use was permitted where you applied, quote the rule in your reply. If it was not, say so plainly, describe exactly what the tool did, and let the university decide.
The countercase: if your use was clearly permitted (UWE, for example, says AI "can help you generate ideas, structure your statement, or help check for readability"), your reply is not a confession. It is a description of permitted use, with the quote. If an agent or consultant prepared your statement, what to do when an agent wrote your SOP covers that case.
Prepare for the interview or writing task
The universities that describe this step describe a live check:
- UBC: applicants "may be required by the graduate program to do so through both written and verbal expression during the admission process."
- Cambridge: "At interview, you should expect to be asked to confirm that you are not using AI."
- UWE: "You will be asked to confirm that you are not using AI during your interview."
Practice for that check, not for a general interview:
- For each paragraph, say aloud in two sentences what it claims and what evidence supports it.
- Pick one technical decision in your project and explain why you chose it over the alternative.
- Explain why this program, naming the course or lab your statement names.
- Write one 150-word paragraph about your project in 20 minutes, without notes. If that is hard, practice until it is not.
UK master's interview questions and PhD interview questions for international students cover the wider interview.
Questions after an AI query
Can a university withdraw an offer I already accepted?
At UBC, yes: the declaration lists "withdraw any offer of admission, whether accepted or not". Cambridge says "If we discover admissions fraud after you've joined the University, we can terminate your studies." UWE, on AI used during an interview, says "If we later find that AI was used, any offer may be withdrawn."
Will other universities find out?
UBC's declaration allows it to "share the information I provided with other post-secondary institutions". Cambridge says "We may disclose application data to third parties when investigating potential fraud." The other pages quoted here do not say either way.
Should I send my version history before they ask?
Not to every program. Send it to the one that asked, with your reply. Keep the files safe until your offers are settled.
Should I email the program again if I hear nothing after replying?
Wait for the timeline the email gave. After that, one short follow-up is reasonable; emailing grad admissions after applying covers when and how.
Get a free review of your statement of purpose
Get a free Quick Review of your statement of purpose against GradPilot's general master's SOP rubric. It reads whether each paragraph carries specific evidence of your preparation, a clear academic goal and fit with the program. PhD applicants can use the general PhD statement of purpose rubric, or start from the PhD SOP review page and choose it there.
More guides for this situation:
- Falsely accused of AI cheating: what to do: your rights and the evidence that clears you.
- Rewrite a ChatGPT SOP in your own words: rebuilding a draft from your own facts.
- All graduate school essay guides.
Sources
Retrieved October 31, 2026, except where noted.
- University of British Columbia, Graduate and Postdoctoral Studies, Use of Generative AI (no year stated): applicant declaration (a) to (e), citation rule, verbal expression, detector guidance.
- University of Alberta, Faculty of Graduate and Postdoctoral Studies, Apply (no year stated): AI prohibition and consequences.
- Technical University of Munich, TUM style guide for application documents (no year stated): AI rule and consequence.
- University of Cambridge, Using AI in your application and Admissions fraud (no year stated): personal statement ban, interview confirmation, fraud consequences.
- UWE Bristol, Using AI in your application (no year stated): permitted uses, interview confirmation.
- Liang, Yuksekgonul, Mao, Wu and Zou, GPT detectors are biased against non-native English writers (arXiv, version 3, July 2023): misclassification of non-native writing.
Get feedback on your statement of purpose
Paste your draft. The free Quick Review shows what to fix against an admissions rubric before you submit.