Should Authors Disclose AI Use? The Honest Case for Yes.

Nobody wants to put a label on their book that says "I had help." It feels like an admission, a chip off the mystique, an invitation for someone to discount your work. So the temptation is to say nothing and let readers assume what they'll assume.

I think that instinct is a mistake, and not a moral one. The case for disclosure is mostly pragmatic. Saying nothing is a bet that the question never comes up, and that bet is getting worse every year. Here's the honest argument for getting ahead of it, plus a way to do it that doesn't require confessing your sins on the cover.

The pragmatic case, not the moral one

Plenty of disclosure arguments lean on guilt: it's the right thing, you owe the reader honesty. Maybe. But you don't need the guilt to land on yes. The self-interested case is strong enough on its own.

Reader trust is your only durable asset. A reader who feels misled doesn't write you an angry email. They quietly stop trusting you and never buy book two. The author-reader relationship runs entirely on the assumption that you're being straight with them. The moment a reader discovers undisclosed AI use after assuming there was none, the damage isn't about AI. It's about the lie of omission. That breach is far more expensive than the disclosure would have been.

The discovery risk only goes up. Right now you can probably get away with silence. But metadata, platform requirements, and reader sophistication are all moving one direction. Betting that nobody ever asks, or that the answer never surfaces, is a bet against time. Front-running the disclosure on your terms beats getting caught flat-footed on someone else's.

Platforms are writing the rules for you. Amazon's KDP already asks whether content is AI-generated. Audible, several trad publishers, and a growing pile of contests now have explicit clauses. The choice between "disclose or don't" is quietly becoming "disclose accurately or violate terms." You'd rather be the author who already had a clean answer ready.

The honest gray areas

A blanket "always disclose everything" rule falls apart on contact with reality, and pretending otherwise just makes writers tune out the whole conversation. The gray areas are real.

Where's the floor? Almost nobody thinks you must disclose spellcheck, grammar suggestions, or thesaurus lookups. Those have been in word processors for decades. But the tools blur upward now. Your grammar checker started suggesting rewrites. Your research assistant started drafting paragraphs. There's no clean pixel where "tool" becomes "co-author."

And there's a genuine fairness problem in vagueness. A label that just says "AI was used" tells a reader nothing useful. It lumps the writer who brainstormed plot options in with the one who generated the whole manuscript, and treats them as the same act. That's not honesty, it's noise. If disclosure is going to mean anything, it has to say what kind. Which points at the solution.

A disclosure spectrum you can actually use

Forget binary. The useful move is to disclose proportionally, matching the statement to what you actually did. Here's a simple tiered version writers can adopt today.

Tier 0: No disclosure needed. Spellcheck, grammar, basic research lookups. Standard tooling. Nobody reasonable expects a note.

Tier 1: "Written with AI assistance." You used AI to brainstorm, get feedback, find plot holes, or get line-level suggestions you chose among. The imagination and the prose are yours; the tool sharpened the process. A line in your acknowledgments covers it. This is the tier most thoughtful AI users actually live in, and it maps cleanly onto using AI as an AI writing coach rather than a generator. We've argued the craft case for why this isn't cheating; disclosure just makes it transparent.

Tier 2: "Co-written with AI." The model generated substantial prose that you revised, but it shaped real passages. The reader deserves to know the words weren't all yours before they form an opinion about your style.

Tier 3: "AI-generated, human-edited." The bulk of the prose came from the model. This is a different product, and selling it as a human-authored novel breaks the implied promise on the cover. Say so.

The point of the tiers is that they let you be honest and fair to yourself. You're not confessing a crime. You're accurately describing a process, which protects you from the writer who did far more AI work claiming the same vague label.

Where to put it (it doesn't have to be loud)

Disclosure does not mean a banner on your cover. The acknowledgments page is the natural home for Tier 1. A single honest line: "This book was written with the help of AI tools for brainstorming and editing; the story and prose are my own." That sentence costs you almost nothing and buys you a clean conscience and a clean record.

For higher tiers, or where a platform asks directly, answer the platform field accurately and consider a copyright-page note. The louder the AI's role in the actual prose, the more visible the disclosure should be. Match the volume to the contribution.

One caution: don't over-disclose to the point of theater. A book that brainstormed a few plot options with a tool does not need a front-cover warning label, and slapping one on just muddies the signal for the books that genuinely do. Proportionality cuts both ways. The reader is best served when the disclosure tells them something true and specific, not when it's loud for the sake of looking virtuous.

The version that's actually good for you

Here's the reframe that makes disclosure feel less like a confession. If your AI use was the good kind, brainstorming, pressure-testing, revision support, the kind that kept you in the driver's seat, then disclosure isn't an admission of weakness. It's a statement of method. "I used a modern tool to work smarter, and the book is still mine." That's a confident position, not a defensive one.

The writers who'll squirm at disclosure are the ones who quietly let the model do the work they're claiming credit for. If that's not you, transparency is a moat, not a liability. It signals to readers that you're the kind of author who's straight with them, which is exactly the trust that survives whatever the AI landscape does next.

Disclose proportionally, put it where it belongs, and stop treating an honest sentence like a scarlet letter. The readers who'd respect you for it are the ones worth keeping.

The cleanest disclosure is the one you barely have to make, because you used AI to improve your own work rather than replace it. That's how Polyz is built. If that's how you want to write, start a free trial.