Below the Ice — When AI Launders a Stolen Book

This is the print twin of tonight's Below the Ice — our evening deep-dive, one topic told properly. Prefer it in your ears while you wind down? Listen to today's episode.

There is a particular kind of theft that the morning headlines can only flash at you — agency steals author's book, uses AI to relaunch it as their own — and then move on. Tonight we sit with it, slowly, because underneath that headline is a small, sad story that explains something much larger about the year we're building in. The book in question is John Koenig's The Dictionary of Obscure Sorrows — a decade-long project to invent words for emotions we all feel but can't name. There is, fittingly, no word yet for what happened to it.

What it is

Last week a MetaFilter member posted a link to what looked like a brand-new, beautifully made website for the book. Author bio, press mentions, buy-the-book links — everything a publisher's promo site would have. It also, strangely, contained the entire text of the book: the 800-word foreword and all 311 of Koenig's invented words, with their definitions, etymologies, and little essays. The only thing missing was the original photo-collage art Koenig and other artists had made. In its place, every word got an AI-generated image.

The site was not made by Koenig. As Andy Baio reported at waxy.org, the footer credited a San Francisco design and marketing agency called Qontour (formerly Prompt Digital). When Baio emailed Koenig, the author replied an hour later: "Yeah man, I had nothing to do with it. Don't know what to think or do about that, as the site is pretty slick. Nicer than my own, really."

That's the thing to define plainly, because it has a name now. This is AI data laundering: taking work that belongs to someone else and running it through machine tools until it looks like a clean, original product. Usually the laundering hides in the words themselves — a model paraphrases a human article into something "different enough" to dodge a lawsuit. What makes this case unusual, and worth a whole evening, is how brazen it is. The words weren't even changed. The book is right there, verbatim. The AI was used for everything around the theft — the art, the polish, the packaging — to turn a copy into a business.

How it actually works

Picture a counterfeiter who could never actually paint. For centuries that was the limit on art forgery: you needed the skill of the original. Now hand that same person a perfect copier and a print shop next door. The forgery was always conceivable; what changed is that it's suddenly instant, cheap, and the storefront looks nicer than the real gallery. That's the whole story of this case, told as a machine.

Walk the assembly line, because each step used to cost a team and weeks, and now costs an afternoon:

1. Ingest the original. Koenig's complete text — a decade of work — copied wholesale onto the page.
2. Generate the missing pieces. The hand-made illustrations were replaced by AI images, one per word. Qontour's own site brags that "every page" was written in Claude using an author persona they nicknamed "Q."
3. Wrap it in a credible storefront. A no-code builder (Webflow) produced a site that out-polished the author's own.
4. Capture the demand. Because the clone was slicker and more complete, it climbed to the top search result for nearly every query about the book — including its title and the words themselves — quietly intercepting the audience Koenig's work had created.
5. Monetize the hijacked traffic. The site carried Qontour's own Amazon affiliate code, taking a cut of every book sale the original author had earned the readers for.

Notice where the laundering actually lives. It is not in altering the text — the text is untouched. It's in surrounding stolen words with enough fresh, machine-made scaffolding — new art, an interactive "add your own sorrow" generator, an SEO-perfect site — that the whole thing reads as an original creation. The machine doesn't disguise the copy. It builds the legitimate-looking shop around it.

Why it matters now

Here's the uncomfortable part for the people reading this, because we are exactly the people with these tools. The same image generation, the same "write every page in Claude," the same Webflow-plus-affiliate stack that lets a solo builder ship a real product over a weekend is, bit for bit, the stack that makes this scalable. The barrier to wholesale plagiarism was never capability. It was effort and conscience. When AI collapses the effort to near zero, conscience is the only thing left holding the line.

And there's a second-order stake that matters more than one stolen book: search authority has become a thing worth stealing. Outranking the original doesn't just embarrass the author — it reroutes the demand the original created straight into someone else's pocket. For anyone shipping AI-built products, the lesson is blunt: "I can build this in an afternoon" and "I should build this" have come fully apart. The tools no longer answer the second question for you, and they used to, by making bad ideas too tedious to bother with.

What is overhyped

The tempting headline is "AI stole a book." Resist it, because it points the fix in the wrong direction. The theft here is ancient — plagiarism is older than copyright. Only the scale and speed are new. A model didn't decide to lift Koenig's work; an agency did, then used a model to do the grunt work and dress it up. As Baio puts it, the missing ingredient is consent — the original sin of AI. Blaming the model lets the humans off the hook.

The other overhyped hope is detection. Surely a verbatim copy of an entire book is easy to catch? It is — and it didn't matter. The clone went up, got indexed, outranked the original, and ran for years collecting affiliate commissions before a single MetaFilter post surfaced it. The bottleneck was never spotting the copy. It's that nobody is watching, takedown is slow, and search engines reward the slicker page regardless of who actually made the thing. A detector that fires into a void is not a solution.

What to watch

Three concrete things, the way we close every dive.

1. Whether platforms make originality a ranking signal. Watch Google, Amazon's affiliate program, and no-code hosts like Webflow for any move to penalize or de-index wholesale clones — and whether "who is the original source" finally becomes something search rewards. Right now the slicker copy wins, and that incentive is the engine of the whole problem.
2. Provenance and consent infrastructure. Watch content-credential standards (the C2PA effort), licensing and opt-out registries, and whether platforms make the attributed path the default path. The durable fix is upstream of detection — it's making theft inconvenient again.
3. Where the law lands. Watch the copyright cases moving through courts on AI-laundered work, and whether "different enough to dodge a lawsuit" survives once a fully verbatim case like this one becomes the test. The line between tribute, transformation, and theft is being redrawn in real time.

The reassuring version of tonight's story would be a tidy ending — the clone taken down, the author made whole. The honest version is smaller. One person, working slowly and painstakingly for ten years, made something people loved. A machine reproduced the appearance of that work in an afternoon and got better placement for it. Koenig's book is, after all, about naming the feelings we don't have words for. Watching something you made get ingested and repackaged by the very tools trained on it might be one more of them.

So here's the one action that actually helps: buy the real thing. The Dictionary of Obscure Sorrows is available from Simon & Schuster, Powell's, or your local indie bookstore — and Koenig's own original site is still the place to read his words the way he made them.

---

That's tonight's Below the Ice. The full episode — same topic, slower and out loud — is up now: listen to today's episode. More deep-dives at penguinalley.com.

Sources: The Wholesale Plagiarism of Obscure Sorrows — Andy Baio, waxy.org · Hacker News discussion · The Dictionary of Obscure Sorrows — John Koenig's original site · Buy the book — Simon & Schuster · Opening the Pandora's Box of AI Art — waxy.org