Authors Guild Tells Publishers Not to Cancel Books on AI Detector Results Alone

What is the Authors Guild asking publishers to do? The Authors Guild has published a set of seven best practices for publishers on AI detectors and suspected AI use in manuscripts. The guidance, AI Best Practices for Publishers: Using AI Detectors and Dealing With Undisclosed AI Use, went up on the Guild's site on Tuesday, October 13, 2026.

The Guild's starting point is blunt. Many authors fear being wrongly accused of using AI by editors, readers and imperfect detection tools, and publishers have already canceled several books for supposedly containing AI-generated text. The Guild already asks authors to be open about their own AI use. Now it wants publishers held to the same standard.

The seven recommendations

1. Put the AI policy in writing. Every publisher should adopt a written policy saying which AI uses it allows and which it bans. That policy should sit in the publishing agreement, or be handed to the author to acknowledge at signing. It should explain what the publisher means by the "originality" and "sole author" promises in the contract, and note that AI-generated text is not original and cannot be copyrighted. The Guild says no publisher should ever cancel a contract over AI use unless the author broke a signed contract or policy that clearly prohibited that specific use, and that term was material to the deal.

2. Don't use AI to edit. Publishers should commit not to run manuscripts through AI for editing beyond simple spelling and grammar fixes. They also should not insert AI-generated content, including artwork, without the author's express written permission.

3. Be open about detectors and internal AI. Publishers should say when and why they use AI detectors, and whether a manuscript might be uploaded to an internal AI system for marketing or sales work. The Guild repeats a point from its April 2026 statement on consumer AI in publishing: staff should never upload a manuscript to a public AI system, and closed in-house systems should be used only with the author's permission. The same goes for detectors. Running a third-party check means sending the text to someone else's AI system, so the publisher should confirm it won't be stored or used for training and should get the author's consent first.

4. Never rely on a detector alone. Even the best detectors make mistakes, and self-reported error rates can't be trusted to hold across different works or over time. The Guild objects to publishers, editors or agents using detector results by themselves to accuse a writer, let alone to cancel a contract. "One false accusation can ruin an author's career," the guidance says. If a detector suggests a policy breach, the publisher has to talk to the author.

5. Train editors and start with a conversation. An editor who suspects AI-generated text may use a reliable, non-infringing detector under the conditions above, then discuss the results with the author and give them a chance to rewrite. If the author won't fix the text, persistent AI use could justify rejecting the manuscript, depending on the contract. Once a book is approved and published, AI use alone shouldn't be grounds for cancellation unless the contract clearly says so.

6. Watch for writers detectors flag unfairly. Many detectors measure how predictable word choice and sentence structure are. That means writers working in a second language, writers in genres with strong conventions, and writers with a deliberately plain style are more likely to get flagged. The Guild says publishers should treat those flags as weak evidence.

7. Don't make authors prove they wrote their books. The Guild opposes any rule forcing authors to use software that tracks or logs their writing process. It does support authors who choose to keep drafts or use such tools on their own, in case they're accused later.

Why it lands now

AI accusations have turned into real disputes this year. PT has covered the Goncourt novel withdrawn over plagiarism and AI findings and the Caine Prize winner who denied using AI. The Guild's document doesn't bind any publisher. It's a template authors and agents can point to when a contract or a suspicion comes up.

What authors can do with it

Ask for the policy before you sign. If a publisher has no written AI policy, ask for one, and ask whether it runs manuscripts through detectors or internal AI tools.

Look at the cancellation clause. The Guild's line is that AI use should only end a contract if a clear, material term bans it. Check whether your contract says that.

Keep your drafts. You don't have to log your writing, but saved versions and notes can help if your work is ever questioned.

Self-publishing has its own rules. Retailers set separate disclosure terms. PT's guide to KDP's AI disclosure rules covers what Amazon asks. The Guild also keeps a companion page of AI best practices for authors.

Sources

• Authors Guild: AI Best Practices for Publishers: Using AI Detectors and Dealing With Undisclosed AI Use (October 13, 2026)
• Authors Guild: Use of Consumer AI Systems in Publishing: Statement and New Model Contract Clauses (April 16, 2026)