AI-Generated Images Allegedly Laundered Into The Dictionary of Obscure Sorrows

A waxy.org investigation published on 20 June 2026 alleges that the original photo-collage illustrations in The Dictionary of Obscure Sorrows were stripped out and replaced wholesale with AI-generated images — a practice the piece characterises as "AI-model laundering" of human-authored creative work.

The Dictionary of Obscure Sorrows is the project of writer John Koenig, built around invented words for emotions the English language has no name for. It began as a web series and YouTube channel before expanding into a published book. The illustrations were not incidental decoration; the photo-collage artwork was part of the original creative identity — assembled by hand, tied to specific emotional concepts Koenig coined.

The waxy.org piece uses the phrase "wholesale plagiarism" deliberately. The allegation is not that an AI model was used to supplement or extend the work, but that existing human-made illustrations were replaced with generated substitutes — a substitution that, if accurate, erases the original creative contribution without attribution or consent.

The term "AI-model laundering" is worth unpacking. It describes a pattern in which AI-generated content is inserted into a pre-existing body of work to pass off machine output as the continuation or equivalent of human-authored material. The mechanism matters here: the source work provides reputational cover and an established audience, while the replacement content sidesteps the creative labour that built both.

This is distinct from generative AI being used to create new works from scratch, where the debate centres on training data provenance and fair use. What the waxy.org allegation describes is closer to substitution fraud — replacing verified human creative output with synthetic material inside an already-published artefact. The legal and ethical surface area is different, and narrower: if the replacement occurred without the original illustrators' knowledge or agreement, it touches copyright, moral rights, and potentially consumer protection, depending on jurisdiction.

For the technology industry, the pattern flagged here is one that practitioners in content integrity and provenance tracking will recognise as an emerging threat vector. Generative image models have reached a quality threshold where substitution inside an existing illustrated work is plausible without casual inspection. The tooling to detect such substitution — C2PA-compliant content credentials, perceptual hashing against known originals, model-output classifiers — exists but is not yet routinely applied to published books or their digital equivalents.

The broader context is one where trust in illustrated and visual creative works is already under pressure. Publishers, platforms, and readers increasingly lack a reliable, low-friction method for asserting that a given image is the one an author or illustrator originally produced. Provenance chains that are obvious in software — commit histories, signed builds — have no direct equivalent in the traditional publishing pipeline.

Koenig's project is a useful case study precisely because it is not a corporate IP portfolio. It is a small, personally identified creative work with a specific cultural following. The reputational harm from substitution, if the allegation holds, falls directly on Koenig and on the unnamed illustrators whose collage work built the book's visual language. That asymmetry — a low-cost substitution with high-cost consequences for individuals — is what makes the pattern worth tracking.

The waxy.org piece does not name the party responsible for the alleged replacement, and the verified facts available at publication do not include a response from a publisher or any other named party. Those are significant gaps. The allegation, as reported, rests on the comparison between original and replacement images, and readers who want to assess the claim directly should consult the source investigation.

What is not in dispute: generative image quality is now sufficient for this kind of substitution to be attempted. Whether it was, and by whom, is a matter the allegation raises but does not fully resolve with the information currently public. The story is worth watching for follow-on statements from Koenig, his publisher, or the illustrators whose work is at the centre of it.