The Macmillan AI policy discussion matters because it connects two author concerns that used to sit in separate conversations: how a manuscript moves through editorial work, and what a publisher may do with that manuscript in AI systems. For emerging writers, the practical issue is not whether every publisher will use the same rules. The issue is whether authors know which questions to ask before copy, catalog copy, reader reports, rights work, translations, and publicity materials begin moving across internal teams.
On May 5, 2026, Macmillan Publishers joined Elsevier, Hachette, McGraw Hill, and Cengage, with author Scott Turow, in filing a class-action lawsuit against Meta and Mark Zuckerberg. The suit alleged that Meta used millions of copyrighted works, including books from those publishers, without permission to train Llama AI models, claiming willful copyright infringement, according to the publishers’ May 2026 press release. That public legal position gives authors a useful reference point: publishers may object to uncompensated AI training by outside technology companies while still setting internal rules for limited AI-assisted publishing tasks.
Why The Macmillan AI policy Matters To Authors
Rights Questions Now Reach The Editorial Desk
Author rights used to be discussed mainly around print, ebook, audio, translation, film, territory, and subsidiary uses. AI has added a different set of concerns. Authors now ask whether unpublished work can be placed into external tools, whether any use of a manuscript could train a model, whether staff can use AI to assess editorial quality, and whether contract language clearly reflects the publisher’s stated practice.
This is where the editorial process becomes part of rights management. A manuscript is not only a creative work; it is also a working document that may pass through editors, production staff, marketers, sales teams, rights teams, and outside partners. If a publisher permits AI tools for some tasks but not others, authors need to understand the boundary. A ban on AI training is meaningful only if the workflow also tells staff what they may upload, summarize, draft, or review.
The Macmillan AI policy And Unpublished Manuscripts
PublishingPost reported that Macmillan circulated a 14-page internal AI use framework in July 2026. The framework allowed employees to use approved large language model tools for limited tasks such as metadata generation, drafting catalog copy, flagging translation review issues, and summarizing rights queries. It also prohibited replacing human reader reports with AI-generated editorial assessments and banned putting unpublished manuscript content into any external AI system, according to PublishingPost’s interview.
The most practical part of the Macmillan AI policy is that it draws a line between administrative support and editorial judgment. Metadata, catalog copy, and rights-query summaries are workflow tools. Reader reports and manuscript assessments involve editorial evaluation of the author’s work. For authors, that difference matters because an AI-assisted workflow can either support human publishing staff or blur responsibility for decisions that affect the book.
Contract Questions Raised By Macmillan AI policy
Policy Statements Are Not The Same As Contract Terms
A publisher’s internal policy can guide staff behavior, but authors often want the same protection stated in the publishing agreement. The research notes show that some authors, including at least three and one with a book on Macmillan’s Fall 2026 list, asked for contractual guarantees that manuscripts would not be used for AI training. PublishingPost reported that Macmillan said it was working on contract language to make that prohibition explicit for authors who wanted it in writing.
For authors, the Macmillan AI policy should prompt a careful but practical contract conversation. This does not mean every writer should demand the same clause or treat every editor as a risk. It means authors should ask whether the agreement, the house policy, and the actual production workflow say the same thing. If the contract is silent while the policy is protective, an author may still want written clarity. If the contract contains broad technology language, an author may want to ask how it applies to AI training, third-party tools, sublicensing, and staff use.
This is not legal advice. Writers should discuss contract language with a qualified agent, publishing attorney, or rights advisor. The author’s job is to identify the question early enough for those advisors to address it before the manuscript is deep into production. A helpful starting point is to compare the policy issue with related contract analysis, such as this site’s discussion of Macmillan AI policy and author contracts.
Questions Authors Can Ask Before Signing
Clear questions make the publishing conversation more productive. They also reduce the chance that an author’s concern will be treated as a vague objection to technology rather than a specific rights and workflow issue.
- Will unpublished manuscript content ever be uploaded to an external AI system?
- Does the contract state that the work will not be used for AI training without express permission?
- Which staff tasks may use approved AI tools, and which tasks require human-only review?
- Can AI be used to draft catalog copy, metadata, or marketing copy, and who approves the final text?
- Will any outside vendor, platform, or service receive manuscript content as part of production?
- How are translation review, rights queries, and publicity materials handled under the policy?
- Who should the author contact if a workflow concern arises after signing?
These questions are direct enough for an editor or agent to route internally. They also help authors separate three issues that are often mixed together: AI training, AI-assisted workflow, and AI-generated publication content.
How AI Rules Change Editorial Workflow
Human Review Remains The Key Protection
The reported Macmillan framework allowed approved AI tools for selected support tasks, but it did not permit AI to replace human reader reports. That distinction matters because reader reports influence acquisition, revision, positioning, and sometimes whether a book moves forward. An author should expect human editorial accountability for that kind of judgment.
Workflow rules also affect timing. If a publisher requires approvals for AI tool use, staff may need extra steps before using those tools on catalog copy, metadata, rights summaries, or translation checks. Authors may not see each approval step, but they should know whether the publisher has a process. A policy without oversight can become informal practice; oversight helps turn policy into repeatable behavior.
Marketing Copy Is Part Of The Same Conversation
Authors sometimes focus only on the manuscript file, but AI policies can also affect promotional materials. Catalog copy, metadata, jacket descriptions, sales notes, rights summaries, and publicity language can shape how a book is presented to booksellers, reviewers, librarians, and readers. If AI helps draft any of those materials, the author’s concern may not be training alone. It may also be accuracy, tone, genre positioning, and whether human staff reviewed the result.
This is why authors should ask to see key public-facing copy at the right stage. The goal is not to control every sentence a publisher uses. The goal is to catch factual errors, misleading positioning, or claims that do not match the book. For writers seeking outside editorial support before submission or publication, utilizing a related resource such as Top Ten Writing Services can aid in framing vendor questions, but publisher-side rights and AI use still need to be handled in the publishing agreement and workflow discussions.
What Emerging Authors Should Do Next
Separate Fear From Process
AI can create understandable anxiety for writers, especially when the discussion involves unpublished manuscripts. A useful author response is to move from fear to process. Ask what the publisher permits. Ask what it forbids. Ask what appears in the contract. Ask who approves tool use. Ask whether staff are barred from placing unpublished manuscripts into external systems. Ask how the policy is communicated to editors, marketers, rights staff, and production teams.
That approach supports a healthier author-publisher relationship. Editors need room to manage schedules, collaborate with colleagues, and prepare a book for the market. Authors need confidence that the work they delivered is not being used in ways they did not approve. A clear policy, matched with clear contract language, helps both sides work with less suspicion.
Keep A Written Record Of AI-Related Answers
Writers should keep written notes of AI-related answers from agents, editors, and publisher contacts. An email confirming that unpublished manuscript content will not be uploaded to external AI systems is easier to rely on than a remembered phone call. If a contract clause is offered, authors should keep the final executed version and any related correspondence in the same file as other rights documents.
Authors should also revisit these questions if the project changes. A new edition, translation, adaptation pitch, teaching version, audiobook workflow, or marketing campaign may involve different teams or vendors. Clear communication early in the process can prevent confusion later, especially when publication schedules become busy.
Macmillan AI Policy And Editorial Workflow Choices
The Macmillan example shows how AI policy now sits inside ordinary publishing work. The public lawsuit against Meta addressed alleged unauthorized training by an outside technology company. The reported internal framework addressed staff use of approved tools for limited publishing tasks. The author contract questions addressed whether protective language should be made explicit in signed agreements.
For emerging writers, the lesson is practical. Do not treat AI policy as a side issue reserved for technology specialists. It affects manuscript handling, editorial assessment, catalog copy, metadata, rights summaries, translation review, and contract negotiation. It also affects author trust. A publisher can have a defensible workflow, but authors still need plain answers before they sign and before their manuscript begins moving through production.
The strongest author position is informed and specific: protect unpublished work, require permission for AI training, preserve human editorial judgment, review public-facing copy for accuracy, and keep contract language aligned with policy. That is how writers can take part in the editorial process without turning every workflow question into a conflict.