Which Grammarly features count as AI?
Before you sort the features, ask the one person who decides. Send this and fill the brackets: "Hi [instructor name], for [course name] I'd like to use Grammarly to fix spelling, grammar and punctuation only, with no text generation or rewriting. Is that allowed for [assignment name]? If it is, do you want me to mention it in the paper?"

Grammarly is AI in the technical sense, because every suggestion comes from a trained model. Schools mean something narrower. They ask whether a tool wrote or rewrote your sentences. Grammarly's own Authorship page splits the product the same way. It counts text "typed and then edited using traditional, non-generative features, such as grammatical error correction, clarity rewrites, and word choice suggestions" as a separate category from text modified with "on-demand AI rephrasing."
| Grammarly feature | Type | What Grammarly's own page says | How Turnitin treats it |
|---|---|---|---|
| Spelling, grammar and punctuation fixes | Correction | Free plan: "Write without mistakes" | Detector "not tuned to target" these edits; in most of Turnitin's tests on human text they were not flagged |
| Clarity rewrites and word choice | Correction (Grammarly's label: non-generative) | Authorship page lists both as traditional features | Not stated on the page I read |
| Tone detection | Analysis | Free plan: "See your writing tone" | Not stated |
| Text generation from a prompt | Generative | Free plan: "Generate text with 100 AI prompts" | Excluded: Turnitin names "draft generation, paraphrasing, summarizing" and says content from these "will likely be flagged" |
| One click full sentence rewrite and tone changes | Generative rewrite | Pro plan: rewriting full sentences with one click | Falls under the generative exclusion |
| AI detector | Detection, writes nothing | Page says "No AI detector is 100% accurate" | Not applicable |
| Authorship reports | Tracking, writes nothing | Support page: tracks the source of text as you enter it | Not applicable |
Grammarly's plans page labels none of these as generative or not, so the Type column for the one click rewrite is my reading, based on the Authorship page counting "on-demand AI rephrasing" as modified text. Turnitin's page names "draft generation, paraphrasing, summarizing" as the generative features and doesn't mention one click sentence rewrites, so that row stays my reading. If a rewrite changes your wording, treat it as generative.
Those prompts are the generative part. Grammarly's plans page shows Free at $0 with "Generate text with 100 AI prompts" and Pro from $12 (the yearly rate; monthly billing costs more) with 2,000 per member per month, rewriting full sentences with one click, and plagiarism and AI text detection. Grammarly pricing breaks down the billing options.
Will Turnitin flag Grammarly?
Turnitin says grammar fixes from Grammarly mostly won't be flagged, and text from Grammarly's generative features probably will. Its AI writing detection FAQ says "Our detector is not tuned to target Grammarly-generated spelling, grammar, and punctuation modifications to content." In Turnitin's tests on human written documents, "changes made by Grammarly (free & premium) and/or other grammar-checking tools were not flagged as AI-written."
The next lines of that FAQ carry the warning. Turnitin adds that the grammar result "excludes content generated by Grammarly's generative AI-powered features," and that "Content produced using these features will likely be flagged as AI-generated by our detector." It also says it has trained its detector to find "leading paraphraser and bypasser tools." So a rewrite tool doesn't make AI text safe. It adds one more thing the detector looks for.
None of this makes a flag impossible. Turnitin tested human documents, not every writer. Grammarly's AI detector page says "No AI detector is 100% accurate" and adds that you should "never rely on the results of an AI detector alone." That cuts both ways: a clean score doesn't prove you followed the rules, and a flag isn't proof you broke them. Your drafts are the better evidence.
Is using Grammarly cheating?
Using Grammarly is cheating only if your course says so, and most syllabi draw the line at tools that write for you. Policies differ by school, by course and even by assignment. A grammar fix that is fine in a history paper can be banned in a writing course that grades your own sentences.
Grammarly's Authorship tool exists because of that gap. Its support page says it "automatically categorizes the source of text as it is entered into the body of a document," so you can show how a document was built. Grammarly's announcement says "Authorship isn't a form of AI detection." It reports where text came from. It doesn't judge it.
Treat the rule as simple. If the assignment says "no AI," skip the generative tools and ask before using the correction tools. If it says "AI allowed with disclosure," use what you like and say what you used. If it says nothing, ask in writing.
How do you check your school's policy on Grammarly?
Read three places in order, then ask a person. The syllabus comes first, because it's the course's own rule. The assignment sheet comes second, since it can be stricter than the syllabus. The school's academic integrity page comes third, and it usually has a general AI line.
These three prompts and messages work in Claude, ChatGPT, Gemini or Whizi. Paste your own text where the brackets are.
- Syllabus reader: "Here is the AI policy text from my syllabus: [paste]. List what I may do with Grammarly, with ChatGPT and with a human tutor. Quote the line that supports each answer. Write 'not stated' wherever the text is silent. Don't guess."
- Question list for office hours: "I use Grammarly for [what you use it for] in [course]. Write five short questions I can ask my instructor about AI and editing tools, from most important to least. Use plain wording and no accusations."
- Written confirmation: "Write a three sentence email to [instructor name] that describes how I plan to use [feature] on [assignment] and asks for a yes or no in writing. Polite, no excuses, under 70 words."
What prompts keep an AI helper to correction only?
A correction only prompt says what the tool may touch and what it may not, then asks for a list instead of a rewrite. Any chatbot, including ChatGPT, Claude and Gemini, will rewrite your whole paragraph if you let it. These four stop that.
- Proofread only: "Proofread the text below. Fix only spelling, punctuation and clear grammar errors. Don't change word choice, sentence order or my voice. Return the corrected text, then a list of every change you made. Text: [paste]"
- Explain, don't fix: "Read my paragraph and list each grammar error with the rule it breaks in one sentence. Don't give a corrected version. I'll fix them myself. Paragraph: [paste]"
- Flag wordy sentences: "Mark any sentence in this draft longer than 35 words. Say why it's hard to follow in under ten words. Don't rewrite any of them. Draft: [paste]"
- Check against my thesis: "My thesis is [thesis]. Here's my draft: [paste]. List the paragraphs that don't support it. Don't suggest new content, only point out the gap."
The first prompt is the closest match to a classic grammar check. The other three keep every sentence yours. If your policy bans all AI help, none of these is allowed either, so check first.
How do you disclose and prove your own work?
Disclose in one or two plain sentences, and keep a record of what changed. A short note takes a minute and protects you if a detector or a reader asks questions later.
- AI use statement: "Write a two sentence AI use statement for my paper. Tool: [Grammarly or other]. Features I used: [spelling and grammar fixes, no text generation]. Say that all ideas and wording are mine. Keep it neutral and factual."
- Change log: "Turn these notes into a table with three columns: tool, feature, what it changed. Notes: [paste what you used and where]. Add no entries I didn't list."
- Version summary: "Here are two versions of my paper, before and after editing: [paste both]. List the kinds of changes between them in five bullets, such as spelling, punctuation or word swaps. Say which changes altered meaning."
- Quiz yourself: "Ask me five questions about my essay [paste] that only its author could answer, one at a time. After each reply, tell me what I missed." If you can answer, you can defend the work in a meeting.
The Authorship feature does part of this for you. Grammarly's support page says paid plans give analytics with "Copied from a source" and "Typed by a Human" labels. In Google Docs, version history also shows when text appeared. Neither replaces a conversation with your instructor.
What mistakes get Grammarly users into trouble?
Most trouble comes from five habits, and every one has a fix.
- Accepting every rewrite in a batch. The fix is to accept suggestions one at a time and read the sentence aloud.
- Using text generation for a first draft and not saying so. The fix is to disclose or skip the feature, whichever your policy requires.
- Treating a clean detector score as permission. The fix is to rely on your drafts and the written policy, not a score.
- Using a rewriter or humanizer on AI text. Turnitin says it detects paraphraser and bypasser tools, and Grammarly's Authorship reports where text came from. The fix is to write the draft yourself or follow your course's AI rules.
- Assuming one course's rule covers every course. The fix is to check each syllabus.
Where does one app for drafting fit in?
If your course allows AI drafting help, one app with several models beats a stack of single purpose tools. Whizi, which publishes this page, runs Claude, GPT and Gemini together. Pro carries all three in one app, and its first week is $1.99.
The free plan gives 10 messages a day on Whizi AI, Auto and Claude Haiku, with no card. Use it to run the correction only prompts above on your own text. It isn't a detector, it doesn't hide AI use, and it can't tell you whether your school allows a tool. That answer comes from your instructor.
Grammarly is still the better pick for in editor spelling and grammar underlines inside Gmail and Docs, where a chatbot has to be pasted into. Its Authorship report is also something Whizi doesn't offer.
What can't this answer tell you?
It can't tell you your school's rule, and it can't promise any detector result. Detectors are probabilistic, vendors update them, and a school can overrule a vendor. The only source that settles your case is your own course.
It also can't cover every Grammarly feature. Grammarly adds tools often, and its plans page doesn't label each one as generative or not. When a new feature writes, rewrites or paraphrases, assume it counts as generative until your instructor says otherwise.
- Copy the exact AI sentences from the syllabus and assignment sheet
- Use correction tools only until your instructor confirms more in writing
- Skip text generation and rewrites unless the course allows AI drafting
- Save version history for every graded paper
- Add a one or two sentence AI use note when the policy asks for it
Frequently asked questions
Does Grammarly count as AI in Turnitin?
Grammarly's grammar fixes mostly don't, but its generative features probably do. Turnitin says its detector is "not tuned to target" Grammarly's spelling, grammar and punctuation edits, and that content from Grammarly's generative tools "will likely be flagged." Check your own course rules too.
Does Grammarly count as generative AI?
Only some features do. Text generation from a prompt, full sentence rewrites and rephrasing are generative. Spelling, grammar and punctuation fixes are not. Grammarly's Authorship page calls grammar correction, clarity rewrites and word choice suggestions "traditional, non-generative features."
Is using Grammarly cheating in college?
It is cheating only when your course bans it. Many syllabi allow grammar and spelling tools and ban tools that write for you. Read the syllabus, then ask your instructor in writing. How to use ChatGPT for homework covers the same line for chatbots.
Can the free version of Grammarly be flagged as AI?
Yes, if you use its generative features. Grammarly Free includes 100 AI prompts a month, and text from those prompts is generative content. The free grammar and spelling checks are the part Turnitin says its detector isn't tuned to flag.