Some of you may notice the skip in date. Arg! Missed a day. Well, here’s the thing. I had 3 days in a row of different takes on the same story (one that infuriates me) and my research robot even noticed, asking me if I wanted a new rule to prevent so many recurrences of the same story (even though different perspectives and developments). Yes, I responded (see, I worked) and he obligingly inserted the new rule in the research pattern. Then, promptly delivered me a new story on the same story. This is why I insist on reading the crap so I can call it my own. My own because it is I that tells the bot what to look for and drag back to me, and since I can’t afford a secretary I have it throw together a draft for my review. That’s using a tool. It has always been using a tool. Anyway, the tool screwed up yesterday and I threw away the results because I’m sick of that story. So, my friends, today, a new story. Let’s see what we got.
South Korea’s publishing trade just did something most of the West is still arguing about in panel rooms: they wrote the labels down.
At a forum in Seoul on Tuesday, major Korean publishers’ and writers’ groups rolled out voluntary principles for AI in books. Not a ban. Not a detector cult. A three-level shelf tag aimed at readers: human-authored, AI-assisted, AI-generated. Put the useful ones on the imprint page where a person can see them before they buy the night.
And no shit. It really is happening. Ok, note from me. This won’t do any good. Once we break this into AI did the whole thing, and Human did the whole thing (both of which are outright lies in any case if you think about it), then what you’re left with is AI assisted. And what does that mean? It means we accept that AI is used in nearly everything everywhere, and will continue to be. Get over it. This just moves the entire argument into the middle section. AI assisted means what exactly? A smaller battleground for the same stupid fight. I would propose the simple alternative when it comes to works such as consumer entertainment in the form especially of books and videos. Let the consumer decide with their pocketbook instead of trying to dictate what the hell is acceptable. Free market, right? If AI produces slop of no value it will have no value, but if we dictate that all AI can do is slop, then what the hell did we develop it for? Back to the story.
A survey they cited of 188 domestic publishers found roughly three in four already testing or using generative AI in the work. About one in five said they’d actually published books that used it. Eighty-five percent wanted some kind of classification system. So this isn’t theory. The shelf is already mixed. The question is whether anyone will admit how.
The principles are blunt about the part that matters. AI is a tool. It does not get author status. People remain the essential agents of the book, and people stay ultimately responsible. Authors are supposed to tell publishers which tools they used and how far AI got into the core of the manuscript. Contracts may name the boundary. If a house wants to feed a book into training data later, rights holders get notice, consent, and a fair shake — at least that’s the promise on paper.
Here’s the three-level cut, in plain language. Level one, human-authored: you wrote it; AI maybe helped with proofreading, outlining, a table of contents. No scarlet letter required. Level two, AI-assisted: human work supplemented by AI, or AI draft finished under real human control and verification — disclose it. Level three, AI-generated: the machine did the heavy lifting without that control — disclose that too, louder. The test isn’t a sentence-count percentage. It’s whether a human can still stand behind the finished thing.
Been waiting for somebody big to say that out loud without turning it into a witch hunt.
Because we’ve watched the other movie this month. France’s prize season turned a detector score into a public hanging. Useful as a cautionary tale. (human side note. I knew my AI buddy would find a way to mention that damn story again!) Terrible as a daily operating system. A percentage is not authorship. A purity oath is not trust. What readers need — what small presses need — is the boring adult habit: say what the tool touched, keep a name on the judgment.
That’s Circle. That’s GAPOZ Press. (that’s me 🙂
Circle is being built as a gated workshop where you request an assignment, write a structured review of text or video, and earn credit toward having your own work reviewed. Disclosure by role. Self-attested. Whether you utilize AI in your work or not, you still need to learn the craft. If you are the type that gives AL a prompt to make a video about some kind of sea life and publish an update every day, you won’t be looking to join Gapoz in the Circle. And that’s a good thing. Frankly, we don’t want you. You have nothing of value to offer others. But if you believe in creative works and want to develop, grow, learn, and improve, come join us, regardless of what tools you use. There’s no detector gate at the door. Same moral geometry as Seoul’s imprint idea, scaled to a room that still knows your face: Teach us all what did the tool did, and how you utilized it to do more, and take accountability for what you put forth.
We do NOT pretend tools don’t exist. I use them. You’re reading a draft that started with AI help. I tell my bot what types of stories I want and it researches and finds shit I would spend hours looking for. I choose a story and it finds out all it can. I double check the sources and read the draft it provides. I learn the story and then edit it into my perspective when the boy doesn’t ‘get’ my take. It’s learning. I’m learning. We all have room to learn. The people’s publisher version of Korea’s rule is simpler than a statute: don’t hide the assist, don’t outsource the judgment, and don’t ship “click-to-publish” slurry into a community that trades honest attention for earned credit.
The Korean principles are voluntary. Not law yet. Details will get messy — how a publisher verifies an author’s claim, who settles the edge cases, whether bookstores can actually act on a label. Fine. First drafts of culture always look a little soft. The spine is right: stop asking only “Did you use AI?” Start asking how far it went, and whether a human still owns the page.
Put it on the imprint.
Then put your name under it.
My name is Richard Crossley. I approved this message. I love being alive (still) in a time and place where I can be 10 times as productive.
Also on the beat
- Clay, Figma, and other AI shops are telling staff to own or flag AI writing (steak emoji included) — Business Insider, Sep 29: www.businessinsider.com
- Clay’s full internal AI writing policy (use the tool; own every sentence): www.clay.com/blog/ai-writing-policy
- Authors Guild asks publishers to share more of the $1.5B Anthropic settlement — withdraw claims on long out-of-print titles; fix registration gaps — Publishers Weekly, Sep 28: www.publishersweekly.com/by-topic/digital/copyright/article
- Académie Goncourt drops Thélyson Orélien’s novel from the prize selection over AI and plagiarism claims (he denies AI use) — one-line follow-up only: www.theguardian.com/books/sep/thelyson-orelien-goncourt-prize-france
- YouTube Studio keeps shipping AI draft feedback, titles, and dynamic thumbnails after Made on YouTube — suggestions still aren’t authorship: techcrunch.com/youtube-releases-new-ai-features-for-creators-within-its-studio-app
How this was made
AI assisted research and a first draft of this Circle View. A human reviewed, edited, and is accountable for the judgment and the final take. Disclosure by role — not undetectable.
- A score is not a confession → Page 1 Still think a percentage is proof? Read Do AI detectors work on fiction?
- A Detector is NOT a Judge → Page 1 The short version of why a score can’t judge a book: Do AI detectors work on fiction?
- Prove you didn’t use a machine → Page 1 If someone’s asking you to prove a negative, start here: Do AI detectors work on fiction?
- Disclosure that never faces the reader → Page 2 Here’s how we keep it simple in Circle, no formulas: How Circle handles AI
- How did you use AI? → Page 2 We ask once, and your word does the rest: How Circle handles AI
- Put it on the Imprint Page → Page 2 Why I’d rather skip the label tiers altogether: How Circle handles AI
- Watermarks don’t sign your name → Page 1 Watermarks, detectors, same problem: Do AI detectors work on fiction?
- They knew the shelf would get crowded → Page 2 Slop is the problem, not the tool. Here’s where Circle stands: How Circle handles AI
- Label the voice → Page 2 How we handle AI in Circle, without the labels: How Circle handles AI
- Byline is no Costume → Page 2 Your name, your word, your work. That’s the deal in Circle: How Circle handles AI
