AI book covers look like a free way to launch a serial cheaply, but on Royal Road they convert worse than a competent budget human cover and filter out the exact genre-savvy early readers who later become paying subscribers. I think paying a real artist $200 to $500 once, then reusing the work across the whole series, is the math that actually adds up for a web fiction author.
There are no universal rules here, and I want to say that up front, because the people who feel strongest about AI covers tend to talk in absolutes and the people who feel strongest the other way tend to dismiss the whole question as moralizing. My own view is somewhere in the middle, but it is closer to "don't use them" than to "use whatever works", and the reason has almost nothing to do with the ethics argument that usually frames the debate. I think the marketing math says skip AI covers for a serial launch, and the ethics math just happens to agree with the marketing math, which is convenient but not what should drive the decision.
Why AI book covers hurt conversion on Royal Road
Royal Road readers can spot an AI cover from the thumbnail row, and the audience most likely to follow a new serial is the one most likely to filter past it. I have spent more time than I would like to admit watching trending rows on Royal Road, and the pattern is consistent in a way that I think a lot of new authors miss. The readers who matter most for a launching serial, the ones who will leave a follow on chapter one and a rating on chapter three and a comment on chapter ten, are not casual readers wandering in from Reddit. They are power readers who have been on the platform for years, who know the names of the cover artists Travis Bagwell and Will Wight worked with, who can tell at a glance whether a cover came out of Midjourney or out of a working illustrator's portfolio, and who treat cover provenance as a credibility signal before they read a single sentence of the blurb.
That is the audience that pays. They are the cohort that converts to Patreon at five percent or higher, the cohort that buys the eventual Kindle release, the cohort that reads the audio when Podium picks the book up. And in my experience, watching how the trending page shakes out over a launch week, that audience filters AI covers out at the thumbnail stage. Not because they have done a careful ethical analysis (some of them have, but most have not), but because the visual fingerprints of an AI cover (melted hands, plastic skin, soft-focus armor that almost looks right, a sword hilt that does not quite connect to its blade, a face symmetrical in the wrong way) read as "amateur" the same way a typo in the first sentence reads as "amateur." It is a shortcut their eyes have learned to make, and they make it before they think about it.
You can argue that this is unfair, and you would not be wrong. There are AI covers that are genuinely well-composed, and there are human-painted covers that look much worse than any Midjourney output. But the reader doing the two-second scan does not care about that distinction, and a launching serial cannot afford to lose the two-second scan. Cradle did not launch with a beautiful cover. Defiance of the Fall did not launch with a beautiful cover. Dungeon Crawler Carl launched with a cover that I think most people would call functional rather than gorgeous. What those books had in common was that the cover read as work, not as a shortcut, and that is the signal you are paying for.
How much a real cover actually costs in 2026
A serviceable original cover from a working illustrator costs between $200 and $700 in 2026, which sounds like a lot until you spread it across the lifetime of a series. I think the price point that worries most new authors is real (rent is rent, day-job hours are day-job hours) but the framing of the decision is usually wrong. The cover is not a one-shot launch expense the way a launch ad spend is. The cover is an asset that anchors every chapter you ever publish, every Patreon tier you ever sell, every Kindle release down the line, and (if things go well) the spine of an Audible page that potentially earns for ten years. Amortize $400 across a hundred-chapter serial and you are paying four dollars per chapter for the single most important piece of marketing copy on the page.
My advice, then, is to think of the cover the same way you would think of buying a good chef's knife if you cook every day. The first time you spend $250 on a knife it feels obscene, because the $30 knife at the supermarket cuts onions just fine. Two years later you notice the $30 knife went dull, got rusty in the dishwasher, and was quietly costing you ten seconds on every onion. The real cover is the chef's knife. The AI cover is the supermarket knife that you eventually replace anyway, except that in the cover case the act of replacement also means losing the brand recognition you built around the original art, which is a problem the kitchen knife does not have.
Reedsy is the market I would point a first-time author at, for the obvious reason that the vetting filter does most of the work. Bluesky is the second-best market, because that is where most of the working SFF cover illustrators have moved and because their portfolios are scrollable in a way that 99designs portfolios are not. Direct outreach (a polite DM, a sentence about the book, a budget number stated up front, a request for a couple of references) gets you a real conversation about ninety percent of the time. The remaining ten percent of artists are simply too busy, and they will tell you so quickly enough.
The ethics question, and why it is not the question that matters first
I think the moral debate about AI book covers is real, but it is not the load-bearing argument for a new web fiction author, because the marketing math says skip them long before the ethics does. The ethics argument is straightforward and not really mine to make: training data, scraped portfolios, displaced illustrators, all of it is true and all of it has been documented well by people who study the question full time. I find the argument persuasive on its own terms. But I notice that when an author who has decided to use an AI cover is being argued at, the ethics frame tends to harden their position rather than soften it, because the ethics frame reads to them as moralizing from someone who is not paying their rent.
That is why I prefer the marketing frame. The marketing frame does not require you to share my (or anyone else's) moral intuitions about training data. It only requires you to look at the conversion rate on Royal Road of two visually comparable books, one with a competent human cover and one with a Midjourney cover, and to notice that the gap is wide enough to matter. My feeling about this is that the gap is wide enough that even an author who is genuinely indifferent to the ethics question, who simply wants to launch the serial that pays for the kid's daycare, should still choose to spend the $300 and skip the AI cover. The ethics happen to agree with this conclusion, which is good, but the conclusion does not depend on the ethics.
Sarah Lin, the author of Street Cultivation and one of the better-organized business operators in the genre, has been public about the fact that she commissioned every cover in her catalog from a working illustrator and that the spend paid for itself many times over. Pirateaba ran The Wandering Inn for years on covers that were drawn by hand, even when AI covers became cheap and available. Andrew Rowe (whose Sufficiently Advanced Magic still sits near the top of the litRPG and progression-adjacent rankings) commissioned a series of painted covers from a single artist precisely because consistency of the visual signature mattered more to him than the per-cover price tag. None of these authors made the choice on moral grounds alone. They made the choice because the math worked out that way.
When a placeholder cover is honest
If you cannot afford a real cover yet, a clean typographic placeholder beats an AI cover almost every time. I think this is the move that gets the least attention in this debate, and it is the move I would recommend to a new author whose entire art budget is zero. A bold serif title in white-on-black, a tagline below it in a contrasting color, the author name at the bottom in something simple, that is a cover that signals confidence and intentionality. Mother of Learning spent years on Royal Road with a cover that was barely more than a piece of text and a stylized circle, and the book became one of the most-read serials in the genre's history. The reader who matters reads "clean and confident" off a typography cover. The reader who matters reads "shortcut" off an AI cover. Those two readings come from the same shelf and they are not the same.
The mini-example I want to leave you with is this. Imagine two covers side by side on the Royal Road trending row, both for a litRPG with a sword-and-numbers premise. The first is a black background, the title set in a heavy gold serif, the author name beneath in plain white, and a single small numeric icon in the corner. The second is a painted AI swordsman in armor that almost looks right, holding a sword that almost connects to its hilt, standing in front of a castle that almost has a consistent perspective. Which thumbnail do you click? The honest answer, if you read in this genre and you have looked at thousands of these thumbnails, is that you click the typography. The typography reads as a writer who has not yet been able to afford a cover but has clearly been able to afford taste. The AI cover reads as a writer who has decided that the cover is not worth the effort, and the reader extends that judgment to the prose before they have read a single word of it.
What changes when you can afford the real cover
Once your serial is earning even a small income, commission the cover you wish you had had at launch and re-list. Royal Road allows cover updates, the algorithm does not punish them, and the act of swapping a placeholder cover for a painted cover is one of the highest-impact promotional moves available to an author who has already built a small following. Patrons love the relaunch energy, the existing reader cohort feels rewarded for being early, and a small re-promotion in the relevant Discords can usually pull in a meaningful second wave of follows. The point of starting with the typographic placeholder is not that you stay there forever. The point is that you do not poison the visual identity of the series at the launch stage with an asset (the AI cover) that you will both want to replace and feel obliged to apologize for once you do replace it.
My closing thought is the boring one and the one I think matters most. A serial succeeds or fails on the slow accumulation of trust over hundreds of chapters, and trust is built out of small decisions made consistently in the same direction. Skipping the AI cover is one of those small decisions. The reader who follows your work for five years is the same reader who would have filtered the AI thumbnail past in the first week. Treat the cover as if that reader is already watching, because, statistically, on Royal Road, they are.
Common questions about web fiction covers and AI art
Are AI book covers banned on Royal Road?
Royal Road does not currently ban AI-generated covers outright, but the moderation team has at various points required disclosure and has pulled covers that contain visible AI artifacts or appear to use other artists' work without permission. The unwritten reader rule is harsher than the written platform rule, and that is the rule I think most authors should plan around. By the time the platform rule is enforced against your book, the reader rule has already lost you the launch.
How much does a custom web fiction cover cost in 2026?
A serviceable original cover from a working illustrator usually runs $200 to $700 in 2026, depending on whether you want figure work, just typography, or a full painted scene. Top-tier serial covers, the ones you see on the front page of Royal Road and on the bestseller list at Aethon, are typically in the $800 to $1,500 range. Reedsy, ArtStation, and direct outreach to artists on Bluesky are the three markets I would actually use.
Will readers really stop reading a story because of the cover?
Most will never start. The cover is doing the click-through work on a Royal Road trending row or a Kindle category page, where a reader is scanning twenty thumbnails in two seconds. A cover that triggers the AI pattern-match gets skipped before the blurb is read, which is a worse failure mode than a clumsy first chapter, because the clumsy first chapter at least gets a chance to be read.
Can I use a free typography-only cover as a placeholder?
Yes, and I think this is one of the most underrated moves a new author can make. A clean, bold, typography-first cover that uses a strong serif title and a single accent color reads as confident and intentional, while an AI cover reads as a shortcut. Pre-fame Cradle and pre-fame Mother of Learning both launched with covers a reader would call basic, and neither held the books back.
What should I look for when hiring a cover artist for a serial?
I look for an illustrator who has done multi-book series work before, because they understand that the cover has to function in a thumbnail row and has to support a recognizable spine when you eventually move to print. I also ask for a layered source file in the contract, because you will inevitably want to swap a tagline, add a series number, or re-promote the book with a slightly different palette six months later, and not owning the layers is a small disaster.
by Jacob Tam · June 9, 2026
I run IlorisNovel, a platform for writers of web fiction. If this kind of craft writing is your thing, the rest of the blog lives here.