What AI Means for Authors (and Publishers)
AI is everywhere — but it does not have to destroy your dream of becoming a traditionally published author.
Traditional publishing has survived major technological shifts before. Meghan Stevenson lived through the rise of e-books and e-readers, when many people predicted that print books and independent bookstores would disappear. That did not happen. E-books took part of the market, but print books remained.
Her message for aspiring authors is clear: traditional publishing will survive AI, readers will still want real books, and authors’ ideas will still matter.
Hosted by Meghan Stevenson, a traditional publishing expert, professional book collaborator, and founder of Meghan Stevenson Books, this episode explores how AI, copyright, intellectual property, and large language models may affect entrepreneurs, experts, creators, and thought leaders who want a traditional nonfiction book deal.
Will AI replace traditional publishing or authors?
No — not according to Meghan’s perspective in this episode.
AI may change how people work, research, and complete low-impact tasks. But Meghan does not believe it can replace the knowledge, judgment, experience, creativity, and human understanding that authors bring to a book.
Traditional publishing also has a financial interest in protecting authors’ copyrights. Publishers invest money, time, and expertise in creating books, paying authors advances, and sharing future earnings. They benefit from authors creating original intellectual property that can be licensed and sold.
For Meghan, the real issue is not whether AI exists. It is how authors choose to use it.
She believes that entrepreneurs, experts, creators, and thought leaders should avoid using AI for anything proprietary or substantive. That includes using a large language model to generate the actual text of a book.
Low-impact uses may be different. Meghan shares examples from her work, including using AI-generated Zoom transcripts as reference materials and using Perplexity to format citations and conduct research. But she distinguishes those activities from generating book text.
Can you use AI to write a traditionally published book?
Meghan’s answer is no.
She first saw an AI clause in a contract in 2024 for a manuscript that was later published by Penguin Random House. The clause stated:
“Writer’s contributions to the Work shall not be generated, whether whole or in part, by the use of artificial intelligence.”
The clause specifically referenced large language models, or LLMs, such as ChatGPT, Claude, and Google’s Gemini.
Meghan carved out exceptions for limited uses of AI in her work process, such as transcript generation, citation formatting, and research support. The agent and author agreed to those specific uses.
But generating text for the book using an LLM was not part of the agreement.
Her reasoning comes back to copyright, intellectual property, and the limitations of current AI outputs.
What does copyright have to do with AI and books?
Copyright is a form of intellectual property that gives creators exclusive legal rights to reproduce, distribute, display, perform, or adapt their original work.
Books are covered by copyright, as are songs, music, videos, photographs, and illustrations. According to Meghan’s explanation, when you create art in the United States, you automatically hold the copyright to that work.
Authors retain copyright even when they receive a traditional book deal.
A traditional publisher does not take away an author’s copyright. Instead, a book deal is a license that allows the publisher to sell the author’s work on their behalf. The publisher helps make the copyright official through the U.S. Copyright Office.
This is important because publishers have a financial interest in protecting copyright. Their business depends on licensing authors’ work, creating books from that work, paying advances, and sharing in the profits.
The exchange is straightforward:
The author creates ideas and intellectual property
The publisher licenses the work
The publisher creates and sells a book
The author receives an advance and may later share in profits
As a professional book collaborator, Meghan’s business model is different. Her contracts allow clients and authors to own the work she creates for them, including the right to reproduce, distribute, display, perform, and adapt it.
How long does copyright last?
Meghan explains that, in general, copyright lasts for the life of the author plus 70 years.
After that period, the work enters the public domain. That is why people can create books based on older works, such as Pride and Prejudice and Zombies.
There is also an exception called fair use, which can allow someone to use a snippet of a book, movie, or song without infringing on the original creator’s copyright.
However, fair use does not have a hard-and-fast definition. It can vary depending on the type of work and the circumstances. Meghan’s example is that fair use involving one Taylor Swift song could be different from fair use involving content from her Eras Tour book.
When copyright is registered through the U.S. Copyright Office, an author or publisher can sue if someone infringes on the work. If the dispute goes to court, a judge may ultimately decide whether the use was fair.
Why is AI a copyright concern for authors?
Meghan’s concern is that the companies behind major AI models used material without permission to train their systems.
According to the episode, companies behind models such as ChatGPT, Claude, Gemini, and Perplexity trained their AI systems using free internet content and public-domain creative work. Meghan also says these companies went further by using pirated versions of books rather than paying copyright holders for their work.
The episode references a lawsuit involving Meta, bestselling author Scott Turow, and major publishers Macmillan and Hachette. The lawsuit alleges that Meta knowingly copied copyrighted materials from pirate websites to train versions of its Llama language model.
Meghan describes this as one example among multiple pending lawsuits involving AI and copyright.
Her concern is that when people use an AI model, they may be using a tool built from the copyrighted work of authors and online creators who were not asked for permission and were not compensated.
For authors, this is not only about legal rights. It is also about creators’ ability to make a living from their work.
How are traditional publishers responding to AI?
Meghan points to specific copyright language from Penguin Random House that appears in its books. The language states that the publisher values copyright because it fuels creativity, encourages diverse voices, promotes free speech, and creates a vibrant culture.
It also asks readers not to reproduce, scan, or distribute any part of the book without permission and states that no part of the book may be used or reproduced for the purpose of training artificial-intelligence technologies or systems.
Meghan says HarperCollins has similar language in its books and expects other publishers to use similar clauses if they are not already doing so.
From her perspective, the publishers’ position is clear: AI should not violate copyright.
Meghan’s other concern is that large language models predict rather than understand.
In the episode, she quotes Perplexity’s explanation that LLMs do not have grounded understanding or beliefs. Instead, they operate on statistical patterns in text, rather than a mental model of the world.
She also quotes the explanation that current AI models do not inherently know whether something is true or false. Unless they are combined with tools or retrieval systems, they rely on patterns from training data and can produce confident but incorrect information.
For Meghan, prediction based on combing the internet does not replace more than 20 years of publishing experience, a human brain, or a real understanding of the industries in which she works.
She is willing to use an LLM for limited, low-impact tasks. But she does not trust AI with important, creative, or proprietary work — not even grammar decisions.
Meghan acknowledges that AI can make people feel as though they must adopt it immediately or risk being left behind. But she points out that many of the people promoting this urgency are making money from others’ use of AI.
She also notes that some people are more concerned than excited about AI’s increased use in daily life.
If you feel hesitant about AI — or nervous about what it may unleash on the world — you are not alone.
You do not have to hand your expertise, creative ideas, book text, or intellectual property to an AI tool simply because AI is everywhere. You can decide what is appropriate for your business, your work, and your values.
Traditional publishing has survived multiple versions of technology. E-readers did not eliminate print books or independent bookstores, and Meghan believes AI will not eliminate authors or traditional publishing.
Readers still want books. They still want authors. And they still want messages that matter.
For entrepreneurs, experts, creators, and thought leaders, your experience, ideas, expertise, and perspective remain valuable. A traditional publisher can still help you understand how to pursue a book deal, develop a proposal that attracts a literary agent and major publisher, and create a manuscript that reaches readers.
AI may be developing quickly, but it does not replace your message.
Questions answered in this episode
Will AI destroy traditional publishing?
Can you still get a traditional book deal in the age of AI?
Can AI replace authors?
Can I use AI to write a traditionally published book?
Can I use ChatGPT, Claude, Gemini, or Perplexity to generate book text?
Can authors use AI-generated transcripts as research materials?
Can I use AI to format citations or conduct research?
What is copyright?
Do authors automatically own copyright to their work?
Do traditional publishers take away an author’s copyright?
What is a traditional book deal?
How long does copyright last?
What happens when a work enters the public domain?
What is fair use?
Why is AI a copyright issue for authors?
How are publishers responding to AI training on books?
Why can large language models hallucinate?
Why does AI sometimes give confident but incorrect answers?
Do I need to use AI to avoid being left behind?
Have a great idea for a book but don't know where to start? MeghanStevenson.com/quiz
