A marketing agency reprinted an entire bestselling book, swapped the art for DALL-E, and outranked the original

What was copied, and how the money actually flows

Andy Baio's waxy.org investigation documents a clean act of wholesale theft. Qontour, a San Francisco web-design agency formerly called Prompt Digital, republished John Koenig's The Dictionary of Obscure Sorrows — a New York Times bestseller, Simon & Schuster, 2021 — in its entirety on a domain one word off from the official one. The complete 800-word foreword, all 311 invented words with their definitions, etymologies and essays: Koenig's decade of work, reproduced. His original photo-collage illustrations were swapped for DALL-E 2 images, a GPT-4 "Submit A Sorrow" generator was added, and the knock-off outranks the real site in search.

The monetization was the part that took tracing. A reader followed the link:

"Just to be clear: The bootleg site is pointing to the Amazon listing of the actual book… The Amazon link was created with Amazon Associates… (you can see the affiliate link code, tag=promptdigital-20…). This is how AI slop can be monetized: poorly gated Amazon programs" — a commenter who traced the affiliate mechanism

So the thief earns a referral cut on the original book while siphoning its traffic. Simon & Schuster filed two DMCA takedowns with Google in July. They didn't work.

Why the takedowns bounced

The legal angle drew the sharpest reversal in the thread — that running text through a model may be enough to defeat the one tool publishers reach for:

"DMCA only applies to 1:1 copy, if you used AI to convert it to something else, then DMCA is the wrong tool… you would have to prove that you can use the said AI tool to create the exact word by word copy and that is going to be very expensive and shaky in court" — a commenter on how AI laundering defeats DMCA

A developer in the thread had lived exactly this:

"I worked on something for 3 years which I give away for free… and a thief took my software, ran it through ai to rebrand everything and relaunched as their own app. Unfortunately the ai missed a few Easter eggs I had hidden so the theft is undeniable. Google and Apple are useless for dmca unless you have a court order." — a developer describing first-hand AI-laundering of his own project

Is this even an AI story?

Not everyone bought the framing, and the dissent is worth keeping:

"some random tiny scam company copied a book without permission in blatant violation of copyright law, and AI was briefly and tangentially involved with the final product." — an R&D specialist pushing back on the AI angle

There's truth in that — plagiarism predates the transformer. But the mechanism is new in a way that matters: AI didn't just copy the book, it made the copy cheap to produce, cheap to disguise from automated takedowns, and trivially monetizable through an affiliate link to the very thing it stole. The old version of this scam needed a human to retype 311 entries and still lost to a DMCA notice. This one doesn't, and didn't.