Editing in the Age of AI Suspicion: Protecting Your Process and Your Reputation
- May 22
- 7 min read
This is a post for editors.
Much to the editorial world’s misfortune, AI has quietly rewritten one of the unspoken rules of our work. For years, when a client got back a clean, consistent, carefully reasoned edit, they assumed a living, breathing, coffee-dependent human had been at the keyboard. Lately? Not so much. More of us are fielding a brand-new question—sometimes asked outright, sometimes just hinted at with a blank stare:
Did you use AI to do this?
If you’ve never used generative AI on a client’s manuscript, that question feels offensive. It pokes at the exact thing we take the most pride in. It did for me—which, full disclosure, is the whole reason this post exists. I’d love for us to get ahead of this together, because the editors who handle it calmly and transparently are the ones who’ll keep their clients’ trust while the ground keeps shifting under all of us.
Here’s how it went down for me. I turned in a long copyedit, felt good about it, and then a few months later when I saw the client in person—whom I’ve worked with before and genuinely like very much—asked me point blank: “We feel you might have used AI to edit that manuscript.” I was speechless. I got angry (but didn’t show it) because my integrity was being questioned. All sort of feelings came rushing up. Because I want to continue working for this client, I kept my business hat firmly on and stated clearly “No, I didn’t use AI. I don’t use AI in any of my client’s work. I would never use AI on my client’s work unless they explicitly ask me to use it. I use AI in my own internal admin processes for running certain parts of my business.” And then the conversation gradually pivoted to something else.
I’ll be honest, though: I didn’t sleep that night. That question sat with me for days. How could they think that? Why would they think that? What in my work raised a flag? What could I have done differently? I thought, and thought, and thought more about it.
So let me clear the air right now:
I’m AI literate—I know what these tools do and, just as importantly, what they don’t.
I do not use AI on any client’s work. Full stop.
I do use AI for internal, behind-the-scenes business tasks when it helps, following a strict protocol I built from the classes I’ve taken and my own ethics.
When I do use it, it’s through closed, paid accounts with the security settings locked down, so none of my writing, prompts, or inputs get fed back to train the models.
All that said, I feel there’s more I need to do now. This whole topic is going to keep evolving, and what we are able to control will keep changing (no one asked us if we wanted to have AI in Adobe Acrobat, MS Word, or our Google searches—the companies just added it in). And to be honest, I’m still irked and haven’t completely gotten over this. But I also understand my client and what they are up against themselves: trying to catch up with technology, copyright laws getting rewritten, writers with manuscripts that need editing, marketing teams that have to grab every viable spot for attention, sales reps that have to generate ROI, plus . . . and everyone in their purview (including competitors) probably using AI too.
So here’s what I’ve found worth thinking through, in case any of it helps you too.
Why even AI-free work can draw suspicion
The cruel irony is: the cleaner and more consistent your work, the more it can read as “suspiciously smooth” to a nervous client. To someone who isn’t trained to spot the difference, polished professional prose and a language model’s output can look like cousins. Throw in the publishing world’s AI apprehension, and a little suspicion is almost baked in—no matter how scrupulous you are.
It also helps to remember that the question “Did you use AI?” is often a stand-in for something else. Sometimes the client really means this didn’t feel as careful as I expected, or I’m not sure I got what I paid for—the integrity question and the quality question get all tangled up together, and the issue may run deeper.
A copyedit is one of the last visible stops on a long production train ride, and when something feels off, it’s awfully tempting for everyone upstream to point at the caboose instead of at their own car. A brief that was never clearly defined, check-ins that didn’t happen, in-house oversight that was thinner than the project needed, earlier editorial stages that got skipped or rushed—it all tends to surface at the copyedit, because that’s where the manuscript finally gets read slowly, line by line. And the copyeditor can end up being held responsible for a gap that opened several cars back.
None of this is a reason to wave off the concern—and it is absolutely not something to say out loud in your own defense, because that just sounds blame-shifting. It’s a reason to respond to the actual situation instead of only the words. Sometimes that means showing the care you put in. Sometimes it means a calm, factual rundown of what the manuscript looked like when it landed in your inbox and what you flagged along the way. Untangling the real issue—and being able to point to your documented process when you do—may settle things a lot faster than defending your honor in the abstract.
Don’t reach for an AI detector!
When you feel accused, the gut instinct is to grab a tool that’ll clear your name. Don’t. AI detectors are not always reliable, they rely on probability and they’re especially trigger-happy with false positives on exactly the kind of clean, conventional prose that good editing produces (especially with nonfiction). A detector can’t prove your innocence, but it can absolutely manufacture doubt that was never there.
Keep building a paper trail of human process
What you can do is make the human fingerprints on your work visible. These are hard to fake and quietly convincing:
Use Tracked changes— which show the back-and-forth, the second-guessing, and the inconsistency of genuine human judgment.
A running style sheet, documenting every call you make as you go: hyphenation, names, timeline, capitalization. Nearly impossible to reverse-engineer, and some of the best proof that a real person crawled through the text.
Author queries and marginal comments, which capture your reasoning in real time.
Version or revision history baked into the document itself.
TextExpander snippets (or screenshots of them) showing the custom abbreviations you use for your longer pre-written notes, links, or images you drop into queries.
A copy of your editing process checklist.
All of this is just good practice anyway. Make it your default on every project and the evidence is already sitting there if anyone ever asks.
Put your policy in writing, up front
The best home for your AI policy is your contract and your proposal—not a panicked afterthought once the question has been asked. A short clause that separates internal business use from the editing of the client’s manuscript does the heavy lifting.
Also, make sure your actual toolkit matches your promise. If you’re claiming the manuscript never touches generative AI, double-check that the shiny new AI features in your grammar checkers and other software are actually switched off. Your promise and your practice have to be synchronous.
If you’re asked directly
Stay non-defensive, even when it stings. Thank the client for raising it (yes, really). Offer to walk them through your process, flipping the conversation from “did you cheat?” to “here’s how I work,” which is a much friendlier. If helpful, loop in the managing editor or project lead. Proactive communication is both reassuring in the moment and protective down the road.
Where I fell short
I reached out to the PM twice with queries—but honestly, I should have done it more. And here’s my biggest unforced error: I needed to stay in my copyediting lane and not start solving or addressing problems that belonged back in the developmental stage. That one’s on me, and I won’t do it again. (It’s the occupational hazard of also being a developmental editor—even when, like on this project, I was not wearing that hat. The urge to fix everything dies hard. Especially if the client didn’t want it fixed or didn’t feel there was an issue and now you’ve created additional work for them.)
Separate the accusation from the feedback
If real errors slipped through, own them gracefully—but keep that separate from the integrity question. Misses happen, especially on a manuscript that showed up less polished than promised. Owning a miss is professional. Letting it curdle into doubt about your honesty is not and smushing the two together don’t help anyone.
The new normal
None of this is going anywhere. Client anxiety and AI suspicion in editing is part of our landscape now, and some version of this conversation will eventually find every working editor. The ones who come out the other side with their reputations intact won’t be the ones who protested the loudest—they’ll be the ones with calm processes, clear contracts, and a visible, documented, unmistakably human way of working, with our without the technology.
And if you’re reading all this thinking it makes perfect sense in theory but you’re not sure how to actually pull it off—how to word that proposal clause, build a process checklist, or hold your nerve when the question lands—please reach out through my contact form and let’s talk. Truly. Bring your questions, your horror stories, your half-finished style sheets. You’re not in this alone.
Linda Ruggeri is a nonfiction bilingual editor and writer based in California who specializes in memoirs, biographies, cookbooks, and Spanish translation reviews. She’s the author of Networking for Writers which received an IBPA Gold Medal in 2025, and The Conference Notebook, an intentional guide for attending and networking at conferences.
Comments