ChatGPT will proofread your draft. Asking it to remove its own habits is asking the source.
ChatGPT is a capable proofreader, and it will rewrite anything you paste. It also answers differently every time, hands back a new draft instead of a list of what was wrong, and reaches for the very phrasing readers have learned to spot. This page covers where each one wins.
| RemoveAISlop | ChatGPT | |
|---|---|---|
| Spelling and grammar | 49,000-word dictionary in US and British English, 35 grammar rules that stay quiet when unsure | Handles context a rule cannot, such as whose versus who's, when you ask it to proofread |
| What you get back | The flagged phrases, each named with the pattern behind it, and a rewrite | A new draft. Ask, and it will describe its edits in its own words |
| Same draft, same result | Always. Rules are deterministic | Different wording on each run |
| AI phrasing | 749 catalogued terms, 30 sentence patterns and 23 formatting habits, flagged by name | Will cut clichés when asked, though its own output is where most of that phrasing comes from |
| Your numbers, names and quotes | Passed through a rewrite unchanged | A rewrite can alter a figure or add detail, so check it against your original |
| Writing new material | None. It edits what you wrote | Drafts, outlines, replies, anything you ask for |
| Does it send your writing to a model? | No. Rules only, with no model client in the product | Yes. Free and Plus chats train OpenAI's models by default unless you switch it off in Data Controls |
| Price | Five drafts a day free. Creator is $7.49 a month billed yearly, or $9.99 monthly | Free with usage limits. Plus is $20 a month |
Where ChatGPT does something better, the row says so. Checked against their public documentation and pricing on the date above.
Try it on your own draft
Frequently Asked Questions
Can I just ask ChatGPT to remove the AI phrasing?
You can, and it will cut the obvious ones you name. The catch is that the replacement wording comes from the same model, so new stock phrases tend to appear where the old ones were, and a second run gives a different result. A fixed catalogue flags the same phrase every time, which is what you want from a final check.
Is ChatGPT better at grammar than this?
On the hard cases, yes. A model reads context, so it handles whose versus who's and agreement across a long clause better than a rule that has to stay silent when a reading is ambiguous. Our rules trade some recall for never telling you a correct sentence is wrong.
Is it safe to paste client work into ChatGPT?
Check your contract first. On Free and Plus, OpenAI uses conversations to train its models by default, and you have to opt out under Settings, then Data Controls. Business and Enterprise plans are excluded by default. Here, the draft is checked by rules and nothing is sent to a model.
Can ChatGPT tell if a draft was written by ChatGPT?
Not reliably, and OpenAI says so. It withdrew its own AI text classifier in July 2023, citing low accuracy. Asking a chatbot whether it wrote something gets you a guess phrased as an answer.
Why use this if ChatGPT is free?
Because you need something that does not write. Draft in ChatGPT if that suits you, then run the result through a check that names every machine-typical phrase, keeps your figures exactly as written, and returns the same verdict on the same draft tomorrow.
Can I use both together?
Yes, and that is the most common setup. Let ChatGPT draft or restructure, then paste its output here before you publish. The flags show which sentences still read as generated, so you edit those by hand rather than asking the model to try again.
One pass, before you publish
Spelling, grammar, readability and AI phrasing in a single check, and the rewrite that fixes them.
Open the studio: 5 free rewrites a day