Skip to content
Article Jane Friedman Aug 2026

Jane Friedman: AI and publishing FAQ for writers

Jane Friedman has spent decades covering the publishing industry — her book The Business of Being a Writer is a standard reference, and her work has appeared in the New York Times, The Atlantic, and NPR. Her AI and publishing FAQ is organized around six areas that writers regularly get wrong or ask about in workshops: what publishers currently require, how AI detection tools work, where copyright law stands, how AI training relates to authors’ rights, what contract clauses to watch for, and how licensing revenue works for writers whose work has been used in AI training.

What makes this useful

The document is explicitly non-advocacy. Friedman is not arguing that AI is good or bad for writers; she is documenting the current state of policy, law, and practice, and flagging where things are still unsettled. That framing makes it more durable than most takes on the topic.

Publisher policies are the section most likely to change rapidly. Friedman describes the current landscape as fragmented: major houses have written AI policies, but enforcement is inconsistent and definitions differ. Some publishers require disclosure of AI use in the writing process; others require that the submitted work be human-authored at the word level. Friedman notes where policies have been updated and how recently.

Detection and disclosure is the most practically fraught area for working writers. AI detection tools have documented false positive rates, and Friedman is direct about this: being flagged does not mean a piece was AI-generated, and passing detection does not mean it was not. She covers what disclosure means across different types of publications — trade journals, literary magazines, commercial fiction — and why the expectations differ.

Copyright covers the current state of US law, including the Copyright Office’s 2024 guidance on AI-generated content. The short version: human-authored work with AI assistance can be registered; work where the creative expression is entirely AI-generated cannot. The edge cases are not resolved, and Friedman flags them as such.

Contract clauses is the section most writers will want to read before signing. She identifies the specific language being added to publishing agreements related to AI training, AI-generated content, and rights to use an author’s work in AI systems. She also flags language that would allow a publisher to use an author’s work to train models — a clause that is increasingly common and easy to miss.

Licensing revenue covers the deals that have been struck between publishers and AI companies to license training data, and what (if anything) flows back to individual authors. The answer varies significantly depending on the contract the author originally signed.

Who this is for

Writers who are actively submitting to publishers, signing contracts, or considering how AI tools fit into their writing process. The FAQ is updated, which makes it worth bookmarking and returning to — Friedman notes the update date at the top, and the publishing AI landscape is changing quickly enough that a six-month-old resource is often out of date.