A coalition of leading ghostwriting professionals just released something the industry has needed for a while: a clear, practical framework for how ghostwriters and their clients should handle AI.
The document, called “AI Guidelines for Ghostwriting,” was released on April 22, 2026, by the Working Group on AI Guidelines for Ghostwriting, convened by Gotham Ghostwriters. It covers everything from disclosure obligations to copyright risk, and it’s worth reading closely if you work in collaborative writing of any kind.
Also on Mediabistro
Why This Moment Matters
Ghostwriting has always operated on trust. A client hands you their ideas, their stories, sometimes their most personal experiences, and you transform them into something publishable. That relationship depends on a shared understanding of how the work gets done. AI has quietly complicated that understanding in ways that most client agreements haven’t caught up with.
Are you using AI for writing? How about an AI grammar tool? A transcription service? A chatbot to brainstorm chapter structure? All of those count. And if your client doesn’t know, you’re operating in a gray area that could create real problems down the line, including legal ones.
The new guidelines try to close that gap by giving writers and clients a shared vocabulary for these conversations. That’s more useful than it might sound. When you’re negotiating a contract, having a clear checklist of AI use categories is a lot more productive than a vague clause saying “no AI.”
The Core Principles
The guidelines open with a statement of principles that sets a useful baseline. A few stand out:
- Writing is a fundamentally human activity. The document doesn’t frame AI as the enemy, but it does insist that human-created writing has qualities AI-generated text lacks: originality, taste, and wit. That’s a clear-eyed position, and it’s one most working writers already know from experience.
- AI is a tool, not a replacement. The guidelines explicitly acknowledge that AI can improve efficiency across a range of writing-related tasks. The concern isn’t AI itself but the absence of clear agreements about how it’s being used.
- Technology doesn’t replace trust. AI detection tools are notoriously unreliable. The guidelines correctly point out that no software can substitute for a genuine, documented agreement between writer and client. This is a profession built on relationships, and that doesn’t change because the tools have changed.
A Tiered Disclosure Framework
The most practically useful part of the guidelines is the disclosure framework, which organizes AI use into three tiers.
The first tier covers utilitarian and administrative uses: transcribing audio interviews, checking spelling and grammar, cleaning up citations, reviewing text for inconsistencies. These are the kinds of AI applications that have been built into everyday writing tools for years, and the guidelines treat them as largely uncontroversial, while still requiring disclosure.
The second tier covers research, analysis, and collaboration: AI-augmented web searches, producing research reports, summarizing papers or recordings, using AI as a thought partner during brainstorming. This is where many writers are already operating, often without formally telling clients. The guidelines ask for transparency here.
The third tier covers generative uses: suggesting titles and headings, generating outlines, drafting text that will later be revised, suggesting reorganization. The guidelines note, pointedly, that generating text that won’t be revised further “is not recommended, due to copyright risk.” More on that in a moment.
The key phrase throughout is that “writers cannot comply with a directive like ‘Do not use AI at all,’ because essential tools, including grammar checkers and web searches, incorporate AI.” Writers and clients need to be specific. Vague contract language isn’t going to protect anyone.
The Copyright Risk Is Real
One of the most consequential sections of the guidelines addresses copyright. The U.S. Copyright Office has ruled that material generated entirely by AI without human authorship is not protected by copyright. That means if a ghostwriter generates final text using an AI tool and doesn’t meaningfully revise it, that text may have no copyright protection at all. For a client paying for a book they intend to publish and own, that’s a serious problem.
The guidelines also flag training risk: by default, material uploaded to public AI models can be used to train those models, which could compromise a client’s confidential information or intellectual property. Writers using AI tools on client projects need to know which settings to use to limit this, and clients need to know their writer is paying attention to it.
There’s also plagiarism risk. AI models generate text pulled from existing sources, often without attribution. A ghostwriter who publishes AI-generated content without carefully checking it could be handing a client plagiarized material. And factual errors, what the field calls “hallucinations,” remain common enough that any AI-surfaced fact needs independent verification before it goes into a manuscript.
Clients Have Responsibilities Too
The guidelines don’t let clients off the hook. If a client provides AI-generated source material to their ghostwriter, they need to say so. That material carries the same factual, copyright, and plagiarism risks as anything else AI generates. A client who hands over an AI-written brain dump without disclosing it is potentially setting their ghostwriter up for a problem they didn’t agree to take on.
The framework treats ghostwriting as a genuine partnership, with obligations running in both directions. That framing is right. The best client-ghostwriter relationships have always worked that way.
What This Means for Your Practice
If you’re a working ghostwriter, the practical takeaway is to use this document as a contract conversation starter. The working group explicitly encourages writers to share the guidelines with clients and use it to negotiate a clear agreement on which AI uses will and won’t be part of a given project. That kind of documented clarity protects everyone.
For what it’s worth, Gotham Ghostwriters CEO Dan Gerstein isn’t worried about demand. If anything, he says AI is creating new clients.
“Over the last year we are encountering a whole new sub-class of prospective clients we refer to as ‘AI refugees’ — non-writer authors who tried to use ChatGPT to write their book and found the results at a minimum unsatisfactory and often unusable,” Gerstein told Mediabistro.
He expects the trend to hold, and points to two forces driving it: the coming glut of AI-generated content will pressure authors to tell original, differentiating stories, and AI turns out to be poorly suited for the core work of ghostwriting.
“A technology that relies on human prompts can’t come up with concepts and book hooks that are original to the author’s unique ideas and life experience. It can’t extract insights and stories that can’t be found in any LLM because they exist only in the author’s head. And perhaps most notably, it can’t win the author’s trust, break down their walls, and get them to be vulnerable and go deep in a way that they would never do on their own.”
If you’re new to ghostwriting and still building your practice, it’s worth grounding yourself in the fundamentals of the craft before layering on AI tools. Mediabistro’s guide on how to become a successful ghostwriter covers the core skills, and there’s also a deeper breakdown of what ghostwriters actually do day to day that’s useful for anyone still figuring out whether this career path fits.
Mediabistro also offers a course specifically on getting started with ghostwriting for those who want structured guidance.
On the business side, the AI conversation overlaps with broader questions about how you price and protect your work. If you’re not sure how to factor AI disclosure requirements into your rate structure or contracts, Mediabistro’s piece on setting your freelance writing rate is a reasonable place to start thinking about it.
An Evolving Document
The working group is explicit that these guidelines will be updated as AI tools and industry needs evolve. That’s the right approach. Anyone who claims to have a final, complete answer to how AI should be used in writing is not paying attention to how fast this area is moving.
What the guidelines get right is the underlying principle: clarity between collaborators is non-negotiable, regardless of the tools either party uses. That’s been true in ghostwriting long before AI existed. The new document just gives the industry a shared framework for applying it to a changed reality.
The full guidelines, including the complete disclosure checklist and guidance on how to use the document in client negotiations, are available at gothamghostwriters.com/ai-guidelines.
Topics:
media-news
