Can You Use AI on an MSW or MPH Essay?

Can You Use AI on an MSW or MPH Essay?

There is no field-wide answer, and anyone who gives you one is guessing. Social work, public health, and counseling programs have written rules pointing in at least four different directions, most programs have written nothing at all, and the centralized application services regulate something else entirely.

So the useful question is not "is this allowed." It is: what does my program's sentence actually regulate, and can I describe what I did in a way I would be comfortable submitting? This article answers that in the order you need it.

For the verbatim policy language itself, program by program, our AI policy reference for MPH, MSW, and counseling programs is the directory. This is the decision guide that sits next to it.

Five things a policy sentence can regulate

Read your program's sentence for its verb, not its tone. Across current official pages, the verbs land in five places, and they are not equivalent:

Those five verbs sort almost every case. "Seek the assistance" is the widest — a tool that only comments on writing you already finished is still assistance you sought. "Crafting" and "enhancing" both reach revision. "Generated" reaches the production of text and, read narrowly, not commentary on text you wrote. And a permission with the word only in it is a ceiling, not a door.

Two of those examples end the question outright. If you are applying to the Tennessee MSSW, the Northern Iowa MSW, or the UAB MPH, do not run those statements through an outside AI review of any kind, including ours.

The line that decides most of the remaining cases

Where a program has not banned assistance, the operative distinction is between help with your drafting and someone or something else's authorship.

The clearest public illustration comes from a program with one of the strictest AI rules in the set. San Francisco State's Department of Counseling states that submitting AI-generated content "may result in your application being disqualified from review," and requires applicants to sign an Affidavit of Authorship. Read the affidavit itself, though, and it defines the standard this way: "'Sole author' means that I have written the Personal Statement and, although I may have received feedback and assistance from others regarding the Statement, it is my work and my story. Any writing, revising and editing of the Statement was done by me."

That is the distinction, written by a program that bans AI outright: feedback and assistance are contemplated; the writing, revising, and editing must be yours. The affidavit also states its own purpose plainly — the statement "is needed to determine the level of graduate writing assistance I may need," which is why the prose has to be representative of you.

DePaul's counseling master's puts the same idea in an attestation you paste into the statement itself: "I attest that this essay is an accurate reflection of my writing ability. This personal statement is my original work and has been written by me." Note what "accurate reflection of my writing ability" regulates — not just authorship, but representativeness. A statement can be your own words and still misrepresent you if the prose on the page is not prose you could produce.

Harvard's Chan School does the public health version without mentioning AI at all: "All materials submitted as a part of your application are expected to be your own work," alongside an instruction to credit any source used, "whether that source is a publication or a conversation." That last clause is broader than most applicants notice, and it covers borrowed sample essays as squarely as it covers anything else.

Three programs, three different postures — all current

The spread is not narrowing. Checked on August 14, 2026:

• A ban plus a signed affidavit. San Francisco State Counseling, above.
• Detection, disclosed up front. Cal State Fullerton's M.S. Counseling admissions page states: "AI and plagiarism-detection software will be used to review submissions. The writing in your statement of purpose should entirely be your own work."
• Disclosure rather than prohibition. The University of North Texas counseling psychology doctoral program writes: "we discourage applicants from using generative AI to brainstorm, write, or edit the goal statement," and then: "If you use generative AI at any point in developing your goal statement, you must include a brief disclosure statement at the end of the document describing how it was used."

On the social work side, MSU Denver's prior-cycle MSW prompts ended with a question rather than a rule: "Briefly describe (1-5 sentences) whether you used Generative AI in the process of writing your personal statement and if so, how you used this tool and the prompts you used." Its 2027 prompts publish when that application opens on September 2, 2026, so that wording must be re-read rather than assumed.

A detection disclosure like Fullerton's deserves one caution in the other direction. An AI-sounding result is a statement about how a passage reads, not proof of how it was made; false positives are mathematically unavoidable and fall hardest on writers whose first language is not English. Knowing a program screens is a reason to write in your own register, not a reason to sand your prose into blandness.

How to write an AI disclosure

If a program asks — and more will — you need a sentence or two that is accurate, specific, and boring. Three constructed examples, at three honest levels of use. None is drawn from a real application.

If you used nothing:

I did not use generative AI at any stage of writing this statement.

If you used it for mechanics only, under a permission like Rutgers's:

I used a generative AI tool once, after finishing my final draft, to check grammar and sentence structure. I did not use it to generate, outline, or rewrite any content, and every sentence in this statement is my own.

If you used it for feedback on your own writing, where that is permitted:

After completing my third draft, I asked a generative AI tool to tell me which paragraphs did not answer the prompt and where my role was unclear. The prompts I used were of the form "which parts of this draft do not answer the question below" and "where am I claiming more responsibility than I describe." I made every resulting change myself and did not paste in any generated text.

Three rules of thumb. Name the stage (before drafting, mid-draft, after completion). Name the function (mechanics, structural feedback, generation). And if the question asks for your prompts — MSU Denver's did — you have to be able to reproduce them, which means deciding now to keep a record rather than reconstructing one later. A prompt history assembled after the fact turns a defensible process into a guess.

Never disclose use that a program has prohibited and call it transparency. A disclosure does not cure a violation; it documents one.

What if you already ran your draft through a chatbot?

This is common and worth answering directly rather than ignoring.

1. Check the rule that governs you. If your program's sentence bans assistance or crafting, stop using the tool on that statement now.
2. Establish whether any generated sentence survives. Read your draft against what you actually wrote. Any phrasing you would not have produced yourself should be rewritten from your own notes, not paraphrased from the output.
3. Write down what you did, while you remember it. Stage, function, and the prompts you used.
4. If a disclosure is required, use the record. If none is required, keep it anyway — it costs nothing and it is the only version of this story you will be able to tell accurately in six months.
5. If you cannot tell what came from where, start the draft again. That is a bad afternoon; an integrity finding is a bad year.

If your experience of a chatbot was that it told you the draft was excellent, the uselessness of that feedback has a name and a mechanism — gradeglazing — and it is a separate problem from the rules question.

Silence is not permission

Most program pages say nothing about AI. That resolves nothing in your favor, and across higher education generally the majority of institutions publish no rule at all.

Two traps sit inside the silence.

The university-versus-program trap. A university's academic-integrity policy is not automatically your graduate program's admissions policy, and programs at one institution routinely publish contradictory rules. Start on the page that carries your essay instructions; that is where every sentence quoted in this article lives.

The platform trap. Applicants keep hunting for the SOPHAS rule or the SocialWorkCAS rule. The centralized services regulate responsibility and honesty, not review. Liaison's Applicant Code of Conduct requires applicants to "accurately represent themselves during the admissions process and to provide CAS and programs with accurate, honest, and complete information," and lists plagiarizing personal statements among prohibited conduct — with sanctions that include revoking an application and notifying participating programs. It contains no mention of artificial intelligence at all. The SOPHAS and GradCAS release statements likewise put full responsibility on you and point at a code of conduct, without publishing an AI permission or an AI ban.

So the platform layer settles the plagiarism question and leaves the AI question to your programs. Read the certification inside your own application before you submit it, and where two rules conflict, follow the stricter one.

Finding your rule in five minutes

1. Open the page that carries your essay instructions, not the university's integrity page.
2. Search it for "AI," "artificial intelligence," "generative," and "ChatGPT."
3. If you find a sentence, identify its verb: generating, crafting, enhancing, seeking assistance, or completing the application.
4. Check whether an attestation or affidavit is required, and read the instrument itself rather than the page describing it — as San Francisco State's affidavit shows, the two can say meaningfully different things.
5. Note whether a disclosure is requested, and start keeping the record it would need.
6. If it is unclear, email the program and work from what they send you. An admissions office's written answer beats any inference, including ours.

Where a rubric review fits, and where it does not

GradPilot reads writing you already wrote. It does not draft, does not rewrite, and does not hand back replacement prose to paste in. What comes back is criterion-based critique tied to your own sentences, plus authenticity signals that describe how the text reads.

Where your program permits review of your own draft, that is what these are for: the MPH personal statement rubric, the MSW personal statement rubric, and the counseling statement of purpose rubric, each run from the graduate statement of purpose review page with your program's exact prompt pasted above the draft. Two Quick Reviews are free each day; your first Full Review is $5, once per account, and Pro is $50 for ten Full Reviews.

Three limits belong in the same breath. Scores are revision signals, not admissions predictions. An AI-sounding flag is a signal to inspect a passage, not proof of how it was written. And no review certifies compliance with any program's rule — that determination is yours, made from your program's current words. Where the rule is a ban, the answer is no, and no feedback is worth an integrity finding.

Related reading

• AI policies for MPH, MSW, and counseling programs — every sentence, quoted verbatim
• What you are actually signing in an AI attestation
• When the same school publishes different AI rules
• Nursing school AI policies — the same treatment for NP, DNP, and accelerated BSN admissions
• MSW personal statement examples, explained
• MPH personal statement examples, explained

Sources

All retrieved August 14, 2026. Program rules change between cycles; confirm your own program's current wording and the certification inside your application before you submit.

• UT Knoxville MSSW personal statement — the assistance ban
• University of Northern Iowa MSW admissions — the "crafting" prohibition
• UAB School of Public Health graduate application — "unenhanced by the use of generative AI"
• Rutgers MSW application process and deadlines — the grammar-and-sentence-structure permission
• MSU Denver 2026 MSW personal statement prompts — the 1-5 sentence disclosure prompt, prior cycle
• MSU Denver 2027 MSW application walkthrough — 2027 prompts publish at the September 2, 2026 open
• San Francisco State Department of Counseling admission instructions — the AI prohibition and the Affidavit of Authorship requirement
• San Francisco State Affidavit of Authorship — the "sole author" definition permitting feedback and assistance
• Cal State Fullerton M.S. Counseling admission — disclosed use of AI and plagiarism detection
• University of North Texas counseling psychology prospective students — the required disclosure statement
• DePaul counseling admission requirements — the authorship attestation
• Harvard T.H. Chan application requirements — the own-work and source-credit rules
• Liaison Applicant Code of Conduct — accuracy obligations, the plagiarism prohibition, and the absence of any AI clause