Confidentiality, rights, and risk

Will my manuscript be kept confidential?

It should be, and the reason to ask is that no licensing body exists to require it. Editing is not a regulated occupation, so confidentiality rests on what an individual editor or company undertakes rather than on a rule anyone enforces.

That makes the wording worth checking. An undertaking that covers the manuscript but not the fact of the engagement leaves an editor free to name the author as a client. An undertaking with no end date behaves differently from one that expires on publication.

Every document a client sends us is treated in confidence. That covers the text, the fact that we were sent it, and anything an author tells us about the book while we have it.

The question worth asking any editor is what happens to the file afterward. Working copies, backups, and email attachments all persist unless somebody deletes them, and an author is entitled to know the practice.

Can I ask for a nondisclosure agreement?

Asking for one is unremarkable, and an editor who has done book work has seen the request before. Authors under contract, authors writing about identifiable people, and authors whose subject is commercially sensitive all have obvious reasons, and an editor asked for one should not treat the request as an accusation.

The practical friction is usually about scope rather than principle. Agreements drafted for corporate work sometimes carry obligations an individual editor cannot meet, such as indemnity clauses with no cap or requirements about data handling that assume a company with an information technology department.

Where an editor declines a specific agreement, the useful next question is which clause is the problem. A refusal to sign anything at all is a different signal from a refusal to accept unlimited liability, and only the first is a reason to look elsewhere.

Small jobs raise a proportionality question. An agreement that takes a lawyer an hour to review sits oddly on a two-hundred-dollar proofread, which is why many editors carry a short standard form of their own.

Does a nondisclosure agreement actually protect anything?

An agreement changes what an author can do afterward rather than preventing anything in advance. A signed agreement gives a contractual claim if the text is disclosed. It does not stop the disclosure, and enforcing it costs money that may exceed what is at stake.

Its more useful function comes earlier. Negotiating one forces both parties to state what counts as confidential, how long the obligation runs, and what happens to the files at the end. Most disputes come from those three things never having been discussed at all.

One clause is worth more than the rest of the document. An agreement that says what happens to the files when the job ends gives an author something concrete to point at, while an agreement that only promises silence leaves every copy where it is.

We are editors and not attorneys. An author with something genuinely sensitive should have the agreement drafted or reviewed by a lawyer rather than downloaded from a template site.

Who owns the edited text?

The author, and copyright law reaches that result rather than editorial custom. Copyright exists automatically from the moment an original work is fixed in a tangible form, and it belongs to the author from that moment.¹

Two doctrines are worth knowing because they are frequently misapplied. A work made for hire belongs to the employer or commissioning party. The Copyright Office is specific that a commissioned work qualifies only if it falls into one of nine listed categories, and only if the parties expressly agree in a written instrument signed by them.² A derivative work is one based on an existing work, and copyright in it covers only the additions, changes, or other new material appearing for the first time.³

An edit does not transfer a book to an editor, and no reputable arrangement suggests otherwise. Where an editor contributes original text rather than corrections, that is a different conversation and it belongs in the agreement before the work starts.

Could my book be used to train anything?

Ask, because the answer varies and nothing forces disclosure. The relevant question is narrow: does the manuscript pass through any third-party service, and what does that service's own terms permit it to do with material uploaded to it.

The legal position is unsettled and the Copyright Office says so. Its report on generative AI training, issued in a pre-publication version in May 2025, declines to give a single answer on whether training on copyrighted works is fair use. It offers instead an analytical framework for identifying the relevant facts and policy considerations.⁴ The report also examines voluntary, compulsory, and extended collective licensing as possible arrangements.⁴

An author who wants a commitment should get it in writing rather than inferring it from silence. An editor unwilling to say what tools touch a file has answered the question.

Will the editor use software on my manuscript, and do I get told?

Software has been part of editing for decades, and the category covers very different things. A spelling checker, a consistency macro that lists every hyphenated compound, and a large language model asked to rewrite a paragraph are not equivalent, and lumping them together makes the question impossible to answer honestly.

The distinction that matters to an author is whether a tool reports or replaces. Tools that find and count leave every decision with the editor. Tools that generate text put words in the book that no person chose.

An author is entitled to ask which category is in use and to say which they will accept. At EditFast the editing is done by a person, and the tools that assist are the reporting kind.

Should an editor disclose the software used on a manuscript?

No professional body we consulted requires it in those terms, so it currently rests on the individual editor. That is an argument for asking rather than an argument that it does not matter.

Disclosure is cheap and its absence is expensive. An author who discovers after publication that passages were machine-generated has a problem with their publisher, potentially with their contract warranties, and certainly with their own sense of the book.

The reasonable standard is the one an author would set if asked in advance. Tools that check and count need no announcement, and anything that writes does.

The same logic applies in the other direction. An author who has used software to draft or revise passages is holding information the editor needs, and withholding it makes the edit worse rather than safer.

Was any part of my book produced with software assistance, and must that be disclosed?

That question now appears on editors' intake forms and increasingly on publishers' submission forms, which is why it is worth answering before it is asked. The obligation, where it exists, comes from the publisher's own requirements rather than from law.

The Copyright Office's work on artificial intelligence is the place to watch, and its May 2025 report is explicit that it offers a framework rather than a resolution.⁴ Anyone stating a settled rule about disclosure requirements is ahead of the published position.

For an editor the practical consequence is different from the legal one. Machine-drafted passages behave differently under a copyedit, with confident sentences that turn out to assert things no source supports, and knowing which passages they are changes how carefully they get checked. Saying so at the start costs an author nothing and improves the edit.