Here’s a quiet fact that keeps getting lost under louder AI fights.
Amazon’s KDP rules ask you to tell Amazon when text, images, or translations were AI-generated. You are not required to tell the reader. The disclosure travels upward into the store’s workflow. It does not have to land on the product page where a human is deciding whether to spend an evening with your book.
YouTube’s GenAI rules are narrower still, and more honest about what they care about. Creators must disclose when AI meaningfully alters or generates photorealistic content that could mislead a viewer about what really happened: a real person saying something they didn’t, a real place that wasn’t filmed, a realistic scene that never occurred. Fair enough. Labels got more prominent this year. Auto-detection got stronger. None of that is nothing.
Then read the exemption list.
YouTube says you do not have to disclose generative AI used as production assistance — including tools that create or improve a video outline, script, thumbnail, title, or infographic. Idea generation: out. Captioning: out. Your own voice clone for a dub: out. The words can be machine-shaped. The face can be human. The label stays dark.
Been there. We already watched that loophole get productized. Ghost-creator networks put paid presenters in front of AI-assisted scripts, rode the algorithm, and only got hit after reporters did the work classifiers couldn’t. Enforcement cited spam and deception, not “the pixels look fake.” The camera proved a body. It did not prove an author.
So the big world is building two different honesty systems at once.
One faces the platform: check a box for Amazon. Toggle an attribute for YouTube Studio. Satisfy the store.
One faces the pixels: if the picture looks real enough to fool someone about reality, label it.
What neither system reliably faces is the reader or viewer as a person owed a straight answer about how the work was made. Script authorship can stay invisible by policy. Book AI can be confessed to the retailer and never confessed to the buyer. Detectors argue percentages. Contracts scramble to catch what forms miss. And people who actually care about writing start asking a worse question than “Did you use AI?” — they start asking whether anyone is accountable for the judgment at all.
Circle is being built for the other direction.
Not undetectable. Not a detector cult. Disclosure by role: what the tool did, what a human decided, and who stands behind the final take. A gated workshop where you request an assignment, write a structured review, earn credit toward having your own work read — with a name attached. That’s the opposite of volume dressed up as a person. Honest attention is scarce. Earned credit is the point. Humans stay accountable when the note is hard.
Platforms may keep optimizing for store compliance and photorealism. Fair. That’s their job.
Ours is smaller and meaner in the good way: if you’re in the room, say what you used, say what you own, and face the people reading you.
And no shit. That really is the work.
Also on the beat
- Publishers Weekly’s mid-September reckoning: ops AI is everywhere, creative lines are messy, and trust with authors is doing more work than detectors — https://www.publishersweekly.com/pw/by-topic/industry-news/publisher-news/article/101215-publishing-s-ai-reckoning.html
- YouTube’s own GenAI disclosure page (read the “don’t have to disclose” list, especially scripts and outlines) — https://support.google.com/youtube/answer/14328491?hl=en
- YouTube Blog: more visible AI labels plus auto-detection for photorealistic synthetic video — https://blog.youtube/news-and-events/improving-ai-labels-viewers-creators/
- CASRAI’s publisher-policy landscape: no AI authors, disclose substantive use, and a Global Reporting Standard still in taxonomy phase through fall 2026 — https://casrai.org/guides/ai-in-manuscripts-publisher-policy-landscape
- KDP content guidelines: AI-generated vs AI-assisted, and who you are required to inform — https://kdp.amazon.com/help/topic/G200672390
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
