Jane Friedman thinks the public AI-writing scandals are about to go quiet. That’s not because nobody’s using the tools. It’s because the screening is moving earlier in the process.
In an Oct 30 essay, she argues that many of the books drawing fire in 2026 were written and bought in 2024 and 2025. That was before agents and publishers started routinely running manuscripts through detectors like Pangram. Now they do. Once everyone with a pile of submissions already knows the score, she says, surprise accusations on social media get rarer. Publishing Perspectives picked up the same point on Nov 2. She adds that writers will adapt and learn what makes a manuscript’s history look “troublesome” and what doesn’t.
Surprise. The industry’s answer to a trust problem is another machine.
Gee, I can hardly wait until the AI gets to the point where it can’t recognize itself. huh?
I’m not anti-tool. I’m openly pro-tool. What I can’t stand is all the theater about labels and detectors. The detectors are AI. Using AI to decide whether you wrote something is a strange loop to build a career on. And the quiet version is worse than the loud one. In the comments on Friedman’s essay, novelist Tracy Cooper-Posey put it plainly. If your own prose sets off a false positive, you may never even hear the no. An agent or editor runs the score, decides the manuscript’s history might get messy, and passes. The scandal didn’t end. It just stopped leaving a paper trail.
That’s the big world, with its scorecards and early filters. In mid-October the Authors Guild put out guidelines telling publishers not to cancel a book on a detector score alone. Publishers also shouldn’t make writers log every keystroke to “prove” they wrote it, and they should talk to the author first. That’s fine as far as it goes. It’s still a world built around proving you’re human to a system that isn’t.
In the little world, Circle is being built on a different bet.
In Circle, members request assignments. Reviews are structured and prompted and submitted on the site. They walk reviewers through different lenses and ask for real feelings. Writers can revise and resubmit. Credit you earn can be spent on your own work or someone else’s. We’re not trying to make the law, and we’re not drawing up formulas for what’s allowed. Before your first submission, we ask you once to be honest about if and how you use AI. It’s part of the membership agreement. Honesty, that’s what we ask. It’s your word and your reputation. Members decide for themselves how they feel about AI in writing or production. The one place AI isn’t allowed is reviewing someone else’s work. We are not interested in how AI feels about it. You can get that from a chatbot. What we want from a review is honest, informed human feedback.
Circle can help someone who writes to be the best, but those aren’t the people it was built for. It’s for writers who want to be their best, the ones who write because they must. AI will probably beat everyone at “best” someday. That won’t matter to those writers. What will matter is whether the work cost them something, or whether they just ran a two-minute thought through a machine and called it a book. Low-effort, money-first slop isn’t “yours” in any way that counts. You know it. I know it. Everyone knows it. Being honest about the tool is what counts. A Pangram percentage isn’t.
And I’ll be straight with you: Circle is new. It has almost no members beyond our test accounts. That’s not a soft sell. It’s an invitation to founding members who want to get to work, who’ll have real influence on how this grows, and who’ll help build toward the vision. The bones are in place. We need builders.
If that’s you, join us and earn your satchel and your home office.
Registration is open at gapoz.com. No fees. Membership to the Circle is limited and subject to approval. Satchels and office space is earned, not given away. Come meet people, make some friends, and grow your craft together.
Your name. Your word. Your work.
Also on the beat
- Surveys from IFRRO, CEATL and AOI: over 90% of respondents to IFRRO, the international group of rights organizations, say generative AI has hurt rightsholders’ earnings. Nearly 30% of translators surveyed by CEATL, the European translators’ council, report losing work to AI. The share of UK illustrators losing commissions to AI rose to 46% in the Association of Illustrators (AOI) survey.
- Coral Hart’s 371 AI books: the South African author who writes with AI claims over a million copies sold and more than $1 million earned in 2026. Her business model is sheer volume (AFP via News24).
- The Authors Guild’s AI best practices for publishers: don’t cancel a book on a detector score alone, don’t demand keystroke logs as proof a human wrote it, and talk to the author first. It came out Oct 13 and is still the industry’s reference point.
- Amicus brief from AAP, STM, IBPA and CCC: in an Oct 29 filing, book and journal publishing groups argued against OpenAI’s and Microsoft’s claim that training on their work is fair use. They say a licensing market already exists and that fair use “adapts—not yields—to technology.”
- Dartmouth Provost Santiago Schnell resigns: a faculty review found he used AI without disclosing it in a Washington Post column and in journal work. The committee relied on his Claude chat logs, not detectors (NHPR and Valley News).
How this was made
AI assisted research and a first draft of this Circle View. Richard reviewed, edited, and is accountable for the judgment and the final take. If there’s a [your name here] in here anywhere, that’s on me.
