The KDP upload screen has a new checkbox this year. Right under your book’s content details, before you can hit publish, there it is: “Yes, some of this content is AI-generated.” You used an AI tool to untangle a plot hole in chapter twelve. You ran your query letter through it for a second opinion. You definitely didn’t let it write your testimony chapter — that part came from three years of praying through a diagnosis you didn’t ask for. So does the box get checked or not?
You close the tab without answering. You’ll come back to it. Except the question doesn’t go away, and this year it has more teeth than it used to.
Table of Contents
Three Signals the Rules Just Changed
For the past couple of years, whether to tell readers about AI involvement in your writing has mostly been a matter of personal conscience — a gray, mushy, do-what-feels-right space. This summer, three separate institutions stopped treating it that way, and they didn’t coordinate with each other to do it.
Amazon quietly made its checkbox mandatory. KDP’s help center now states plainly that authors must “inform us of AI-generated content (text, images, or translations) when you publish a new book or make edits to and republish an existing book.” Skip it, and you’re violating KDP’s terms, which can get a book pulled or an account suspended under an enforcement policy that’s already active.
The European Union moved from guideline to law. As of August 2, 2026, Article 50 of the EU AI Act legally requires that AI-generated text on matters of public interest be labeled when it lacks human review or editorial control, and that people be told when they’re interacting with an AI system. This is binding regulation, not a platform’s house rule, and it reaches any provider whose output lands in front of EU readers — including you, if you sell books online.
And a flagship of evangelical publishing drew its own line for entirely theological reasons. The Gospel Coalition’s editorial team published guidelines barring AI from both the writing and editing of anything they publish, stating they want readers to trust that what they read at TGC is “human-made, human-edited, human-assessed, made in human community for other humans.” No regulator required that of them. They chose it, and named the reason: love for the people reading.
Three institutions, with three completely different sets of incentives (a retail platform protecting its catalog, a government protecting consumers, a ministry protecting its witness), landed on the same conclusion within weeks of each other. That kind of convergence, across three unrelated institutions in a matter of weeks, is a sign the ground is actually shifting under how readers will expect this handled going forward, and it landed while most working authors were still treating it as optional.
Where the Actual Line Is
Before you can decide what integrity requires of you, you need to know what the rules themselves actually distinguish, because it’s narrower than the anxiety around it suggests.
Amazon’s own policy draws a real line between two things: AI-assisted work, where you wrote it and used a tool to brainstorm, check grammar, or refine something already yours, and AI-generated work, where a tool produced the actual text, image, or translation, even if you edited it afterward. Only the second category requires disclosure. Running spellcheck doesn’t make you a fraud. Asking a tool to draft your protagonist’s confrontation scene while you tweak the dialogue after crosses the line the moment the tool wrote the words, whatever you did to it next.
That distinction matters: most of what working writers already do (outlining help, research summaries, a second set of eyes on a query letter) sits on the “assisted” side of a line regulators and platforms have now drawn for you. The real anxiety in that KDP tab is simpler and harder: have you been honest with yourself about which side of the line a given chapter actually falls on?
The Question Underneath the Checkbox
Here’s what a compliance checkbox can’t ask you, and what your conscience has to: not “will this get flagged,” but “am I telling my readers the truth about who made these words.”
Scripture doesn’t treat that question as new. “Therefore, having put away falsehood, let each one of you speak the truth with his neighbor, for we are members one of another” (Ephesians 4:25). Paul was writing to a church learning to live honestly with each other in the smallest daily exchanges, long before publishing existed as a category. But a reader who buys your devotional and assumes every word came from your own wrestling with God is your neighbor, in exactly the sense Paul means. Falsehood by omission is still falsehood.
Proverbs is blunter about the stakes: “Lying lips are an abomination to the Lord, but those who act faithfully are his delight” (Proverbs 12:22). Notice what’s paired against the lie — not perfection, not talent, but faithfulness. A writer who discloses an AI-assisted research chapter plainly is closer to that delight than one who produces flawless, undisclosed prose while quietly hoping no one asks.
This isn’t a new argument for this site — the case that AI can’t replicate your particular, lived voice still holds. That argument was always about trust more than detection: whether you can be trusted to say plainly what’s yours and what isn’t, regardless of whether a platform’s algorithm would ever catch the difference.
A Disclosure Practice for the Gray Areas No Rule Reaches
The checkbox only covers what you upload to Amazon. It says nothing about the newsletter you send every Tuesday, the blog post you drafted with help last month, or the devotional excerpt you’re about to post on Instagram. None of those have a mandatory checkbox. All of them have a reader on the other end who deserves the same honesty. Build your own practice before a platform builds one for you:
1. Write your own bright line, in your own words. Don’t borrow Amazon’s or the EU’s definition wholesale; decide, specifically, what “I wrote this” means to you. Most writers land somewhere close to KDP’s assisted/generated split; some hold a stricter line than the law requires. Either is defensible. Having no line at all, deciding case by case, under pressure, on deadline, is the one indefensible option.
2. Put it somewhere a reader can actually find it. A single sentence in your newsletter’s About page or a footer line on your site, something like “I use AI tools for research and editing, never to write the words themselves,” costs you nothing and answers the question before anyone has to ask it. Silence reads as evasion even when nothing dishonest happened.
3. Disclose the borderline cases too. The temptation is to disclose only when a rule forces you to and stay quiet everywhere else. Faithfulness in the small case is the actual test: the newsletter intro nobody’s checking as much as the book upload someone might audit.
4. Revisit the line as the tools change. What counted as “obviously fine” eighteen months ago (basic grammar tools) and what counts as a live ethical question now (a tool drafting a full scene) will keep shifting. A disclosure practice that was honest a year ago can go stale without you noticing.
If you’re navigating a platform terms update or a self-publishing decision alongside this, you’re not the first person to feel like the ground keeps shifting under self-publishing platforms. You don’t have to sort through it alone. CWS’s publishing services exist partly for exactly this kind of moment, when the rules changed and you need a second set of eyes before you commit.
When Honesty Costs You Something
Say the honest thing plainly: disclosure can cost you. A reader who sees “AI-assisted research” on a chapter might trust you less than they would have trusted your silence. A competitor who doesn’t disclose looks more effortlessly prolific than you do. There’s no guarantee that telling the truth here gets rewarded quickly, or ever, in a way you can measure.
That’s the same math behind every newsletter swap you’d think twice about recommending. The honest answer sometimes costs you in the short term and pays in the only currency that actually compounds: a reader who has learned, across enough small honest moments, that you’ll tell them the truth even when silence would have been easier and no one would have known. That reputation doesn’t build overnight, and it can’t be faked by a well-worded policy page. It gets built one disclosed chapter at a time.
Your Next Upload
Go back to that KDP tab, or open your newsletter draft, whichever is actually sitting open on your desktop right now. Before you check the box or leave it blank, ask the smaller, harder question first: does this reader know the truth about how these words got made?
If the answer is yes, publish in peace. If it’s not quite yes, that’s not a reason to panic — it’s a reason to write one honest sentence somewhere your readers will see it, today, before the next upload makes the question harder to answer than it is right now. The line was always going to move again. Your integrity doesn’t have to wait for the next version of the rule to catch up.