How to Protect an Unpublished Manuscript

The fear most new authors have — that someone will steal their book — is almost entirely misplaced. The risks that actually cost authors money and work are duller and more likely, and nobody warns them about those.

So this article does two things: dispenses with the myths quickly, then deals with the real exposures.

---

What you already have

In the US, and in every country signed up to the Berne Convention, copyright exists the moment a work is fixed in tangible form. You typed it; it's yours. There is no application, no fee, no waiting period, and no symbol you have to include.

That means the following are all unnecessary as a creation step:

• Registering with anyone before you can claim ownership
• Adding a © notice (useful as a deterrent and a signal, not a requirement)
• Any third-party "copyright protection" service that charges to timestamp your file

"Poor man's copyright" does not work. Mailing yourself a sealed copy of your manuscript creates no legal right you don't already have, and is not accepted as a substitute for registration in a US court. It's a persistent myth that has survived decades of being debunked. The US Copyright Office says so explicitly.

What registration actually buys you

If copyright is automatic, registration is not pointless — it's just doing something narrower than people assume. In the United States:

1. You cannot sue for infringement of a US work until the work is registered. This is the big one. Automatic copyright without registration is a right you can't enforce in federal court.
2. Statutory damages and attorney's fees are only available if the work was registered before the infringement began, or within three months of first publication. This is why the timing matters enormously. Without it, you're limited to proving actual damages — which for most books is a small and hard-to-prove number, often less than the cost of the lawsuit.
3. It creates a public record, which has evidentiary weight.

Practical rule: you generally don't need to register a manuscript while it's still an unpublished draft. Register within three months of publication. That's the window that preserves the remedies worth having.

Fees, US, as of August 2026: $45 for the Single Application (one work, one author, not made for hire, same author and claimant) and $65 for the Standard Application.

One thing to watch: the Copyright Office has proposed eliminating the Single Application and raising the Standard Application to $85. The proposal went to Congress in July 2026 and is not yet in effect. If you were planning to register several backlist titles, doing it under the current schedule may be worth several hundred dollars.

Outside the US, registration systems vary and are generally less consequential — many countries have no registration at all, because Berne protection is automatic and enforceable without it.

---

The risks that are actually real

Now the useful part. In descending order of how likely they are to happen to you.

1. You lose the file

This is, by an enormous margin, the most common way authors lose a book. Not theft — hardware failure, accidental overwrite, a sync conflict that silently resolved the wrong way, a phone in a river.

The fix is the same as it has always been, and skipping it is a choice:

• Three copies, on two different kinds of media, with one off-site
• Encrypt the off-site copy yourself, before it leaves your machine, so that solving the loss problem doesn't create an exposure problem
• Test a restore once. An untested backup is a rumor
• Keep a copy of the final file corresponding to each published edition, not just the working draft. Reconstructing which draft became the paperback three years later is miserable

2. You lose track of which version is real

Less dramatic, more frequent. Four files named some variation of final, and no reliable way to tell which one went to the formatter. This costs hours and occasionally causes a wrong upload.

The discipline that works isn't folder structure. It's a "current" flag — one file per role is the live one, everything else is history, and the flag is recorded somewhere other than the filename.

3. A collaborator's copy leaks

Editors, beta readers, sensitivity readers, cover designers, formatters, ARC readers. Every one of them gets a copy, and each copy is outside your control.

Proportionate measures:

• A short confidentiality clause in your agreement with paid professionals. Most freelance editors expect this and many propose it themselves. A one-paragraph NDA is normal and not insulting
• Watermark ARCs with the recipient's name or a unique code. This is standard practice and it works — mainly as a deterrent, because everyone knows the copy is traceable
• Use a delivery service rather than email attachments for ARC distribution, so you can revoke access and see what happened
• Don't send the production file when a PDF will do

4. Terms of service you didn't read

Every cloud tool you paste the manuscript into has a section describing what rights you grant. Usually reasonable, occasionally broad. This is covered in detail in How to Keep Your Manuscript Private in the Age of AI — the short version is to read the section called Your Content or License, and to look specifically for sublicensable, derivative works, and language about improving or developing products.

5. Submission exposure

If you're querying agents or submitting to publishers, your manuscript is going to strangers by design. Reasonable precautions:

• Research the agency. Legitimate agents don't charge reading fees
• Send what's asked for and no more — most requests are for a partial
• Keep a submission log with dates and what was sent. If a dispute ever arises, the contemporaneous record is what helps you
• Don't post substantial excerpts publicly if you intend to sell first serial rights

6. Actual theft

Last, because it is last. Wholesale manuscript theft by another author is rare, hard to monetize, and easy to detect once the work is published. The genuine version of this problem is piracy after publication, which is a different issue with different responses (DMCA notices, and mostly, perspective).

---

What to do this month

A realistic sequence, in order of return per hour spent:

1. Fix backups. Three copies, two media, one off-site and encrypted. Test a restore.
2. Flag the current version of every manuscript and every cover file you have.
3. Add a confidentiality clause to your standard agreement with paid collaborators.
4. Read the content section of the terms for the two tools your draft actually lives in.
5. Register copyright on anything published in the last three months, and consider registering the backlist before the fee schedule changes.

None of that requires buying anything.

---

Where AuthorAZ fits

AuthorAZ is made by the author of this site.

Nothing above is solved by software, with one partial exception: keeping the record of which version is current, alongside the covers, editions, ISBNs and registration numbers for each title. AuthorAZ holds that on the device, with encrypted export for your own backups. It doesn't back up your drafts for you and it isn't a substitute for the three-copy rule.

---

Related reading

• How to Keep Your Manuscript Private in the Age of AI
• ISBN, Imprints, Copyright: The Boring Profit Protectors
• Submitting to Agents (The Hybrid Portfolio)

Related answers

• Do authors need to register copyright?
• Should authors use AI?

Sources

• US Copyright Office, Circular 1 (Copyright Basics) — automatic protection, the registration prerequisite for suit, and the "poor man's copyright" myth.
• 17 U.S.C. §412 — statutory damages and attorney's fees; the three-month window.
• US Copyright Office, Fees and the 2026 proposed fee schedule submitted to Congress, July 2026.

General information, not legal advice. US-focused; other jurisdictions differ. Verified 9 August 2026.