The Grammarly Question No College Will Answer (2026)

The Grammarly Question No College Will Answer (2026)

Thirty million people use Grammarly. One U.S. university will tell you whether you can.

That's not a hot take. It's what we found when we read every published AI admissions policy across 174 U.S. universities. Sixty-four schools name ChatGPT in their policy text. Eight name Turnitin. Exactly one — the University of Michigan — names Grammarly. And the one school that names it does so inside a graduate-program policy that simultaneously permits grammar review and prohibits AI drafting, leaving applicants to figure out where their actual workflow lands.

Meanwhile, students are using Grammarly every day to write the very essays these policies are about. The asymmetry between user base and policy coverage is the single biggest practical gap in college admissions guidance in 2026.

This piece is the answer to four questions students keep asking us and getting nothing back from admissions offices:

• Is Grammarly allowed in college essays?
• Does Grammarly count as AI?
• Can I use Grammarly on my personal statement?
• What if my college doesn't say anything about it?

We'll walk through what Grammarly actually is in 2026, what the data shows about which schools have addressed it, and how to map the five permission levels (L0–L4) we use in our policy methodology onto an actual Grammarly workflow.

The headline finding: 1 of 174

Across the full GradPilot dataset of admissions AI policies — 174 schools, every classification at High or Medium confidence — only one institution names Grammarly in its admissions policy text: the University of Michigan. The other 173 say nothing about Grammarly specifically.

For comparison, here's how the major tool names show up in policy text:

Think about that. ChatGPT shows up by name 64 times. The tool students are actually using to polish their essays — the one with ~30 million users and a sentence-by-sentence presence in their writing process — shows up once. Universities have learned to say the word "ChatGPT." They have not learned to say the word "Grammarly."

What Grammarly actually is in 2026

Part of what makes this gap hard to navigate is that "Grammarly" is no longer one product. The same brand spans three very different things:

1. Free tier. Spelling, grammar, basic punctuation, tone signals. Roughly what built-in spellcheck has done since Word 95, plus some style nudges. On Grammarly's own pricing page, this is described as "Write without spelling and grammar mistakes."
2. Pro tier. Adds full-sentence rewrites, paraphrasing, tone adjustment, English fluency suggestions, and "unlimited personalized suggestions." This is the tier where Grammarly stops being a checker and starts being an editor.
3. Generative AI ("Go," AI Agents, Docs). Grammarly markets a generative AI assistant that, per the company, can "generate one-click drafts to help you start faster, build outlines to organize ideas clearly, rewrite sentences for flow, clarity, and engagement, and adjust tone and word choice for any audience." It will also draft full paragraphs from a prompt. The free tier currently includes 100 AI prompts per month; Pro includes 2,000.

Grammarly itself describes its product as AI on a page literally titled "AI at Grammarly," with the line "AI that transforms how people communicate." The company isn't hiding it. The product has migrated, very publicly, from spellcheck into the same generative AI category as ChatGPT.

The problem is the brand didn't migrate. To a student, "I used Grammarly" still sounds like "I ran spellcheck." To an admissions reader, in 2026, it could mean any of the three tiers above — including the one that drafts paragraphs.

The University of Michigan: one quote, three different rules

So what does the only school that names Grammarly actually say?

Michigan is interesting because the undergraduate policy doesn't name Grammarly. The undergraduate admissions site adopts the Common App's fraud policy:

"The University of Michigan expects all applicants to be familiar with and abide by the Common App Fraud Policy."

"The Policy also prohibits the use of AI systems to generate content."

— admissions.umich.edu/apply

That's a generic ban on AI-generated content. It doesn't mention Grammarly by name, and Grammarly's free spellcheck is plainly not what "generate content" means.

The Grammarly mention lives one level deeper, in Michigan's Rackham Graduate School policy for grad applicants:

"GenAI tools, like Grammarly, may be used to review grammar and spelling of your application essays."

But the same Rackham page immediately adds:

"Your application essays should reflect your unique academic, research, and life experiences, and you should be the sole author of all written passages in your essays."

And then it requires every applicant to attest to sole authorship, with revocation consequences if the attestation turns out to be false.

In our rubric, Rackham's posture comes out as L3/D3/E2 — brainstorming and mechanics are fine, but no AI rewriting, AI-specific attestation required, and program-level enforcement teeth. Michigan's undergraduate posture, by contrast, comes out as L4/D0/E0 — a banned posture inherited from the Common App fraud policy, no AI-specific disclosure form, no stated enforcement step.

So the one school that names Grammarly doesn't have a single Grammarly answer. It has three: Rackham permits the spellcheck use case with strings attached, the Law School separately tells applicants "applicants ought not use ChatGPT or other generative artificial intelligence tools as part of their drafting process," and the Ross School of Business goes the other direction entirely, allowing AI essay assistance with mandatory APA citation. That's four different policies under one university brand. (We wrote about exactly this kind of intra-university split in Same School, Different AI Rules.)

The headline still holds, though: across 174 institutions, this is the only admissions policy text where the word "Grammarly" appears at all.

The L0–L4 framework, applied to Grammarly

GradPilot classifies every school on a five-level permission scale (L0 to L4) plus disclosure (D0–D3) and enforcement (E0–E3). You can read the full definitions on the methodology page. For Grammarly specifically, here's how each level translates into practice:

L0 (Silent — 122 schools, 70% of the dataset)

The school has no admissions-specific AI policy. Stanford, MIT, Penn, Princeton, Notre Dame, Rice, NYU, and ~115 other institutions all live here.

What it means for Grammarly: Implicitly fine. The default backstop is the Common App fraud certification, which prohibits generated content but says nothing about spellcheck or grammar review. If you ran free-tier Grammarly across your essay, no one at an L0 school will plausibly call that a violation. Pro-tier sentence rewrites are a grayer zone — defensible, but if you got asked, you'd want to be honest about it. The generative AI / drafting features are over the line by the fraud-certification standard regardless of school silence.

L1 (Permissive — 2 schools)

Generative AI text may be included, subject to fraud and disclosure rules. This is a vanishing tier — only Wisconsin-Madison and Duke currently sit here in our dataset.

What it means for Grammarly: Use whatever tier you want, just don't fabricate experiences. Even AI drafts are tolerated at this permission level, provided the content is true and you take responsibility for it.

L2 (Substantive assistance — 28 schools)

AI may paraphrase, suggest line edits, rephrase sentences, translate the applicant's own words, and provide style/clarity suggestions. The final wording must be the applicant's. Schools like UCLA, UC Berkeley, Caltech, Carnegie Mellon, and Georgia Tech sit here.

What it means for Grammarly: This is exactly Grammarly Pro's job description. "Line edits and clarity suggestions" is what Grammarly Premium does. By the rubric, L2 schools should permit Grammarly Pro use. Note that 6 of these 28 schools (Caltech, UC Berkeley, UCLA, UC Santa Cruz, UC Santa Barbara, UVA) require an AI-specific attestation on top — D3 — so even where the use is permitted, the school may make you affirm it. The generative drafting features ("write a paragraph about why I want to study CS") still fall outside this permission.

L3 (Brainstorm and mechanics only — 14 schools)

AI is allowed for brainstorming, outlining, topic discovery, and basic mechanics (spelling/grammar). No AI rewriting of sentences. No generated phrases or passages in the submission. Cornell, Johns Hopkins, Dartmouth, Northwestern, Tufts, and Rackham (Michigan grad) live here.

What it means for Grammarly: Free-tier spellcheck and grammar review are fine. Pro-tier sentence rewrites are explicitly out — that's exactly what L3 disallows. Generative features are also out. If you're at an L3 school and you accept a Grammarly Pro suggestion that rewrites your sentence, you're outside the line the school drew, even if Grammarly's brand makes the action feel innocuous.

L4 (Prohibited — 8 schools)

No AI use for any part of admissions writing. Brown, BYU, Georgetown, Harvard, SMU, Villanova, Wesleyan, and undergrad Michigan all live here.

What it means for Grammarly: This is the hardest cell. The L4 schools haven't told you "Grammarly is banned" — only Michigan has even named it, and Michigan's L4 mention is for the undergraduate policy that inherits the Common App fraud language. A defensible read is that free-tier spellcheck (which doesn't generate content) survives even an L4 prohibition because the Common App's bright line is about generating content, not correcting spelling. A more conservative read is that "no AI use in any form" means no Grammarly at any tier. We'd default to the conservative read for L4 schools and let the school clarify if they want spellcheck excluded.

"Tools like Grammarly are fine" — the phrase that doesn't appear anywhere

Here is the strange part. In our rubric, the example phrase for an L2 classification is "Tools like Grammarly are fine; don't let AI write it for you." That sentence is the textbook L2 statement. It should appear in dozens of policies. It appears in essentially none.

The reason matters. Universities know what Grammarly is. They know students use it. They could resolve 90% of student anxiety by adding one sentence to their admissions page: Free-tier grammar and spellcheck tools, including Grammarly, are permitted. Generative features that draft sentences or paragraphs are not.

Almost no one has. Instead, the policy literature has converged on a single proxy term — "generative AI" — that has to do simultaneous duty for:

• ChatGPT, Claude, and Gemini (named or unnamed)
• Grammarly's draft generation features (almost never named)
• Grammarly's sentence rewrites (Pro tier; ambiguous)
• Grammarly's spellcheck (free tier; clearly permitted)
• Microsoft Editor, Apple's built-in writing tools, and every other ambient writing assistant

"Generative AI" is a category that no longer maps cleanly to a product. Students can't act on it without making a brand-tier judgment call the school refuses to make on their behalf.

So, can you use Grammarly on your college application?

Here is the closest thing to a clean answer we can give based on the actual policy landscape:

Free-tier spellcheck and grammar review. Use it. At L0 silent schools, it's covered by the Common App fraud floor (which targets generation, not correction). At L1, L2, and L3 schools, it's explicitly within the permitted zone. At L4 schools, it's a defensible gray area, and you can choose your risk tolerance. There is essentially no realistic scenario where a college rescinds an offer because you ran spellcheck.

Premium / Pro features that rewrite sentences. Defensible at L0, L1, and L2 schools. Outside the line at L3 schools (which explicitly prohibit AI rewriting). Off the table at L4 schools. Worth knowing: 6 of the L2 schools require an AI-specific attestation, so even where the use is permitted, you may be asked to affirm it on the application.

Generative / draft-paragraph features. Avoid for any application essay, at any school. This is the use that the Common App fraud policy targets directly. The same Grammarly button that drafts a marketing email for you is the button that crosses the line on a personal statement. Brand familiarity with Grammarly does not change what the underlying action is.

The deeper recommendation: know which tier of Grammarly you actually have running. If you're not sure whether you accepted a free-tier punctuation fix or a Pro-tier sentence rewrite, you're not sure what the school would think you did. Grammarly itself has made that distinction blurry in 2026 — every suggestion shows up in the same purple bar — and the only way to stay on the right side of any school's policy is to know what your own software just did.

What this means for the rest of admissions

The Grammarly gap is the most visible example of a bigger pattern. Related reading:

• The Common App AI policy paradox — the certification 18% of schools defer to is stricter than most students realize.
• Do colleges actually use AI detectors? — what schools check for vs. what they tell you they check for.
• Can you use ChatGPT on a medical school application? — the same question under a stricter policy regime.

You can also browse the full AI policies directory school by school, and the methodology page breaks down how each L/D/E classification is decided.

The honest summary: 30 million people use a writing tool that one American university is willing to talk about. Until higher education catches up to its own student body, you'll have to make the call yourself — and the best version of that call is one where you know which feature of Grammarly you used, on which school's essay, under which level of permission. That's the asymmetry. Now you have a map.