If you use AI editing tools to correct your writing, does it mean it could be flagged as AI-generated?
That’s a question many writers are now asking as AI becomes part of everyday correcting, editing, and proofreading.
The answer isn’t straightforward because correcting writing, rewriting text, and generating content are three different processes.
In practice, basic grammar and spelling corrections don’t replace your writing or change its originality. However, AI detection tools are inconsistent at best, which can create uncertainty and confusion about how writing is classified.
What counts as AI-generated writing?
I’m sure you’ve noticed that many writing sites and services are now asking you to indicate if you have used AI in your work.
Only yesterday, I noticed a new AI disclaimer on Pinterest.
But for authors, Amazon KDP asks if your work was AI-generated or AI-assisted. Its definition of “assisted” is telling: “If you created the content yourself, and used AI-based tools to edit, refine, error-check, or otherwise improve that content (whether text or images), then it is considered AI-assisted.”
The terms edit, refine, error-check, and otherwise improve suggest references to AI editing and correction tools.
What isn’t explicitly addressed is whether non-AI grammar tools fall under the same classification.
This could mean that if you used a grammar checker that does not use AI, you may not have to declare it, which raises the question of whether these tools fall outside that definition.
For authors, it is not yet clear how this information might affect the visibility or treatment of books in the future.
While the obligation to declare AI use in writing is appearing on more and more sites, it’s not always easy for writers to tick the right box.
If a text is AI-generated, yes, the answer is simple.
But for AI-assisted writing, a lot of gray areas remain.
No AI use, AI-assisted, or AI-generated?
AI is being used more and more in all walks of life. But for readers, writers, and publishers, AI is a stigma, even though it is now widely used.
But where do you draw the line between human writing and assisted writing?
Before AI, there was no mark against a writer’s name for using grammar, punctuation, and spelling checkers.
The change in perception occurred when many of these tools added AI features that could do far more than simply correct.
However, not all writers use or even want these extra capabilities. They might use a correction tool with AI, but don’t use these extras.
To try to draw the line, the best way to think about this is what an AI tool changed in your text, and how much it changed it.
Here are some common tasks and how they could be flagged.
Spell checker and typos: Changes and corrections in spelling or fixing typos that make no difference to the original words.
Usually, no AI use
Grammar correction: If the corrections are basic and don’t change the word order, the text retains the words as written by the author.
Probably no AI use
Rewording a sentence: Because of the change in the order of the original words, it is a rewrite, and therefore an assisted process.
Likely to be AI-assisted
Rewriting a paragraph: This involves much more change than at the sentence level, so it is clearly an AI rewrite.
Usually AI-assisted
Generating a paragraph from a prompt: Text generated entirely from an AI prompt does not originate with the writer, so it is generally considered AI-generated rather than human-written.
Usually AI-generated
Generating an entire article: If a first draft originates from AI rather than the author, most publishers and platforms would classify it as AI-generated even if it has been human-edited.
Typically AI-generated
While AI detection tools might come to different conclusions, these examples are a practical way for writers to think about where editing ends and AI-generated content begins.
Staying on the right side of the line
If you are concerned about AI use in your writing or having to tick disclaimer boxes, you can relax a little.
AI is everywhere now and changing day by day, which makes it difficult to avoid.
If you use a Google or Bing search for simple research, you’re using AI. You might also use ChatGPT, Gemini, or Claude to brainstorm ideas and concepts.
In many respects, these tasks are little different from using reference books, traditional search engines, or writing prompts.
Using AI for research is normal now and will not have any direct effect on your writing output.
Where you do need to be aware and careful is when using AI to change, manipulate, or expand on your original words.
It might be subtle, or only a few extra words. But every word you allow AI to write for you increases the AI contribution in the final text.
To stay on the safe side, try to avoid using any AI editing tools that do anything more than correct basic grammar and spelling mistakes.
Also, avoid the process of humanizing an AI draft text. No matter how much you edit it, trying to make it more human, the AI original base still remains.
For many platforms and publishers (like Amazon KDP), that distinction is very important now.
Conclusion
It’s only natural that writers are concerned about the effects of AI on all forms of writing and publishing.
Writers are on the front line and are being affected in many ways. These include challenges ranging from AI content copying to declining readership caused by AI-powered search.
And now the added task of deciding how to respond to AI disclaimers places more pressure on writers to respond correctly.
But questions like “Did you use AI?” are not always easy to answer honestly with a simple yes or no.
Nor is it clear what the consequences of those answers might be in the future. This is perhaps the greatest uncertainty of all.
For now, the most practical approach is to understand how you created your writing and where AI played a role in that process.
If you need to disclose your AI use, the most reliable answer is still the simplest one. Describe your process as accurately as you can, based on what you actually did, and read and understand the terms of the request.
As I highlighted with Amazon, the terms of a request can be very broad.
Author’s note and admissions
Did I use AI to help me write this article?
I did, yes, and here are the boxes I would need to tick.
YES: I brainstormed ideas with Google search and ChatGPT.
YES: I created a basic outline for this article with AI.
NO: I didn’t create an AI article draft to edit.
YES: I wrote this article in my own words.
NO: I didn’t use an AI editing tool to check my writing. (I used the basic ProWritingAid grammar checker.)
NO: I didn’t use an AI style tool to check my writing.
YES: I used AI to check if I had covered all relevant points and if I had made any incorrect assumptions.
From my admissions above, I’ll let you be the judge of how this article should be classified. Is it no AI, AI-assisted, or AI-generated?
Related Reading: Why I Chose To Keep My Blog Despite AI
