Nearly Half of Published Authors Use AI and Most of Them Won't Admit It

I posted that as a note on Substack. 178 likes, 91 comments, and 12 restacks. One person told me I should list the LLM (the AI language model) as the author and put myself in the acknowledgments as “prompt creator.” Another said it’s “false or illegitimate” to use AI for writing. Period.

The objections people have to AI-assisted writing have a lot more to do with ego than they do about writing. The fight is over who gets to call themselves a WRITER.

Every Creative Field Has Had This Fight (And the Gatekeepers Lost Every Time)

Music producers went through this exact thing when digital audio workstations replaced tape recording. “Real musicians record live.” The old guard said Pro Tools was cheating and auto-tune was an insult to talent. The whole idea that anyone could make an album now was treated like a threat to the craft. And then Billie Eilish recorded an album that won five Grammys in her brother’s bedroom using Logic Pro and a USB microphone. Nobody argued about whether the tools were legitimate after that. The industry moved on. The gatekeepers found something else to complain about.

Architects went through it when CAD replaced hand drafting. “Real architects draw by hand.” There were entire professional organizations that resisted the shift. They said the soul of design was in the hand-drawn line, that something essential was lost when you let software do the calculations. Every new building you’ve walked into in the last 20 years was designed in CAD. And the buildings got more ambitious, not less.

This is what happens every single time.

A tool arrives that makes the work faster and more accessible. The people who built their careers doing it the hard way feel the ground shift. They frame the new tool as cheating because that’s easier than admitting the game changed. They gatekeep for as long as they can. And then the tool becomes standard and nobody remembers the argument.

Writing is going through this right now. AI is the tool. The arguments are identical. “Real writers don’t need it.” “You’re not a real author if AI had any part in it.”

The premise of the objections are the same. It’s a different decade and a different tool, but it’s the same motivation and fear of losing relevance underneath. It’s being driven by the same ego.

But the gatekeepers are going to lose this one too.

What the Loudest Critics Are Actually Protecting

Some numbers first so you understand what’s actually happening in the field.

45% of authors surveyed by BookBub (over 1,200 of them) are already using AI in their workflow. Publishers Weekly found a 31% productivity increase among authors who use it. The global market for AI writing tools is projected to hit $47 billion by 2034.

Here’s the biggest and most telling stat of them all. Of the 45% using AI in their writing process, 74% of them don’t tell their readers about it.

Nearly half of working authors are already doing this. Most of them quietly. (Which tells you everything about the climate right now.) And the reason they’re quiet is because admitting you use AI feels like professional suicide because most of the publishing houses and authors associations are taking a hard line on AI generated content right now. In early 2026, Hachette pulled a contracted horror novel over suspected AI use. The Authors Guild launched a “Human Authored” certification program. And virtue-signaling writers are putting anti-AI badges on their profiles like it’s a political campaign.

The writers who are most vocal about AI-assisted writing being illegitimate are almost always writers who built their audience on craft. They spent years developing their skill and earned their position through a process that was slow, difficult, and exclusive to people who had the time, the discipline, and the cognitive wiring to do it the traditional way.

AI changed the economics of that overnight. And that scares them.

A person who thinks beautifully but writes slowly can now produce finished work. Ministry leaders who preach with fire on Sunday but can’t get a blog post written during the week can finally get their ideas into a format that reaches people beyond the room. People who were locked out of the writing world because their brains didn’t work the “right” way can now participate.

That means more competition and more voices in the space that weren’t there before. And if your entire identity is built on being one of the few people who could do this well, that’s a problem for you.

I clearly have strong feelings about this. Wrapping your competitive anxiety in a moral argument about the “sanctity of writing” is ridiculousness. If your writing is good, AI-assisted writers entering the space shouldn’t threaten you. Period. And if it does threaten you, it’s worth asking yourself if your real value was actually the writing itself or just the fact that fewer people could do it and it made you feel special.

The Blank Page Bottleneck

I’ve talked about how my brain works in previous articles, but here’s a recap in case you missed it. I process ideas fast. Frameworks form fully in my head and I can effortlessly explain them out loud and have someone walk away understanding completely. Everything breaks down because my brain is thinking three ideas ahead of the paragraph I’m trying to finish so I get frustrated and then just stop.

I spent YEARS thinking that was a discipline problem. It wasn’t. And once I figured out why, everything changed.

Writing involves two completely different processes. Getting ideas onto a blank page from nothing (generating) and shaping a draft that already exists into something good (editing). My brain is wired for developing ideas quickly and also for editing. I work best when I can explain via talking (typically in an audio to text transcribing tool). Those ideas are usually delivered in a way I would explain it to someone and not necessarily in linear steps or in a format that makes sense simply converted to writing. The beauty of working with AI is that it can take my ideas and thoughts, organize them and get them onto the page for me. Then I can push back, cut, restructure, and refine until it sounds exactly like me. I’m fast at that part and I genuinely enjoy it. Generating the first draft is where my brain stalls, and it always has. The ideas are there. Getting them into linear written form from scratch without help is the part that breaks down.

AI eliminated the blank page. It gives me a draft to edit. And editing is where my actual skill lives. The blank page was just a bottleneck that had nothing to do with whether I had something worth saying or not.

I’m not the only person this is true for. In a world of increasing neurodiversity, the cut and paste playbook no longer works.

The question the anti-AI crowd never answers is this: if the thinking is original, the voice is real, the editing is rigorous, and the final product is indistinguishable from hand-written work, what exactly was lost by not staring at a blank page for four hours first? Nobody has given me a good answer to that.

The closest thing to a real argument I’ve heard is the idea that the struggle of writing IS the thinking. That when you’re fighting to find the right word or restructuring a paragraph for the fourth time or deleting everything because the argument just isn’t working, that friction is what sharpens the ideas. And that if you bypass all of that with AI, the thinking never fully develops.

And look, I take that seriously. And for people who type a topic into ChatGPT and publish whatever comes back, the criticism holds. That work IS shallow because the thinking was shallow.

But in my process, the thinking happens BEFORE the draft. I know what I’m arguing, who I’m talking to, and what I want the reader to understand before Claude writes a word. Then the draft comes back and I fight with it. Sections get restructured. Entire paragraphs get rejected. I rewrite openings three and four times and add the examples and personal details that make the piece mine. The struggle is still there. I’m just struggling with a draft instead of a blank page. The intellectual work is all still happening. The only thing that changed is that the production bottleneck is gone, and the thinking can actually reach the reader now.

If the struggle itself was the point of writing, we’d still be chiseling stone tablets.

For these people, AI is an accessibility tool. Calling it cheating is the same as calling a wheelchair cheating because the person using it can’t walk as fast as you can. And I don’t hear anyone in the anti-AI writing crowd getting riled up over that! LOL.

See for Yourself

This is from this article. The opening of the Blank Page Bottleneck section.

What Claude wrote:

“I’ve talked about my own brain before. I process ideas fast. Frameworks form fully in my head. I can explain them out loud and have someone walk away understanding completely. But sitting down to write them in linear, long-form prose? My brain is three ideas ahead of the paragraph I’m trying to finish before I even get through the first section.

I spent YEARS thinking that was a discipline problem. It wasn’t. My brain just works faster than the traditional writing format allows for. AI solved that by giving me a structured draft I could edit instead of a blank page I had to fill. Editing is a completely different cognitive task than generating from scratch. My brain can edit all day.”

What I published:

“I’ve talked about how my brain works in previous articles, but here’s a recap in case you missed it. I process ideas fast. Frameworks form fully in my head and I can effortlessly explain them out loud and have someone walk away understanding completely. Everything breaks down because my brain is thinking three ideas ahead of the paragraph I’m trying to finish so I get frustrated and then just stop.

I spent YEARS thinking that was a discipline problem. It wasn’t. And once I figured out why, everything changed.

Writing involves two completely different processes. Getting ideas onto a blank page from nothing (generating) and shaping a draft that already exists into something good (editing). My brain is wired for developing ideas quickly and also for editing. I work best when I can explain via talking (typically in an audio to text transcribing tool). Those ideas are usually delivered in a way I would explain it to someone and not necessarily in linear steps or in a format that makes sense simply converted to writing. The beauty of working with AI is that it can take my ideas and thoughts, organize them and get them onto the page for me.”

Claude’s version was competent but generic. And Claude only produced that because I had already loaded it with my voice sample, my guardrails, and detailed context about who I am and how I write. Without all of that, the output would have been generic slop. My version added the frustration (”so I get frustrated and then just stop”), the audio-to-text detail, the distinction between generating and editing as separate processes, and the explanation of HOW AI actually fits into my workflow. That’s what editing a draft looks like.

“Just Make Sure the Author Is the LLM”

Someone in my comments told me that if I write a book using AI, I should list the LLM as the author and put myself in the acknowledgments section as “prompt creator.” Aren’t keyboard warriors fun?

That comment reveals exactly how little most people understand about what AI-assisted writing actually involves.

A “prompt creator” types “write me a blog post about productivity” and copies what comes back. That’s what this person thinks I do. That’s rubbish. Here’s what actually happens when I write.

Before AI writes a single word for me, I’ve already decided what I’m writing about, who it’s for, what problem I’m addressing, what point I’m making, and what I want the reader to walk away thinking. That’s the work, and AI cannot do it for me. If I skip that step, the output is garbage every time. I’ve proven that to myself more times than I want to admit.

Then there’s the system I built. A voice sample I created by recording myself talking for ten minutes and transcribing it so Claude knows how I actually sound. A guardrails document with 98 specific patterns that AI must avoid. Things like mirrored contrast phrases and performed vulnerability, plus dozens of structural and tonal tells I identified by spending months being annoyed by the output and writing down exactly what was wrong with it. On top of that, there are voice rules that define how my content should sound, a personal lexicon of words and phrases I favor and ones I refuse to use, and a set of required patterns that every piece of content has to meet before it publishes. That system took a long time to build and I refine it constantly.

When Claude gives me a first draft, I read every word and push back on phrasing that doesn’t sound like something I’d say out loud. Entire sections get rejected. Then I add the personal stories, the real examples, and the specific details that only exist in my head. The piece goes through a multi-pass editing process that catches AI patterns at the sentence level. And I read the final version out loud to make sure it sounds like me.

I have 98 editorial rules, a voice training system, a multi-pass audit process, and the final creative authority over every word that publishes under my name. The person who left that comment has a snarky one-liner and zero understanding of the process they’re criticizing.

The Part I Don’t Talk About Enough

My process isn’t perfect. If I only talk about the wins, I’m doing the same thing the AI slop creators do, just from the other direction.

Early on, before the guardrails were tight, I published pieces that had AI fingerprints I didn’t catch. Phrases that were too polished, structures that were too symmetrical. I look back at some of those early articles and I can see it. A reader who knew what to look for could see it too.

That’s actually why the guardrails document exists. Every pattern in that document started as a mistake I made. Something Claude produced that I let through because I was tired, or moving too fast, or hadn’t trained my eye to catch it yet. The 98 patterns didn’t appear overnight. They were built from failure, one cringe at a time. (And there were a LOT of cringes.)

If you’re going to use AI to write, you WILL publish things you wish you hadn’t. You’ll miss patterns. You’ll let something through that doesn’t sound like you and someone will notice. The question is whether you learn from it and tighten your system or just keep pretending the raw output is fine.

I tightened. And I keep tightening. That’s the goal.

What I’m Doing Next

I’m writing a book right now using this exact process. Same Claude projects, same guardrails and voice training, same editorial standards I use for every Substack article. And I’m documenting the whole thing as I go.

I’m putting together a working experience where you write your own book alongside me, using my guardrails and my process, and come out the other side with real progress on your manuscript. More details on this are coming soon.

If that interests you, make sure you’re subscribed so you don’t miss it or send me a DM and let me know you are interested.

Write well. Be honest about how you do it. And stop letting people who’ve never examined their own process tell you yours doesn’t count.

---

I’ve written about AI-assisted writing before. If you missed those articles:

What AI-Assisted Writing Actually Is (And Why the Debate Around It Is Missing the Point)

I Have Hundreds of Thousands of Words Written and Nothing to Show for It (Until AI)

If you want the system behind everything I write, the AI Writing Guardrails is the exact framework. 98 patterns to avoid, voice rules, a guided builder for creating your own custom rules, and a 5-pass editing process.

Because “make it sound more human” was never going to work.

Self-Guided or Custom Build options available.

guardrails.aiwithleah.com/info