Manuscript confidentiality for reviewers

When an author shares an unpublished manuscript with you, they're trusting you with work that may represent months or years of effort, and which hasn't yet been publicly timestamped on arXiv or anywhere else. Protecting that trust is one of the most important things a reviewer or endorserEndorserAn arXiv author who is eligible to endorse in a subject area, which depends on their own recent papers there. On ResearchGuild, endorsers are volunteers who opt in to read abstracts. does.

ResearchGuild is independent and not affiliated with arXiv. This article combines arXiv's guidance for endorsers with how ResearchGuild handles manuscripts.

What arXiv asks

arXiv's endorsementEndorsementA statement, made on arXiv's own form by an established arXiv author, that you are someone who can reasonably contribute to a subject area. It is not peer review and doesn't guarantee acceptance. page is direct: if you're sent a paper to read as part of an endorsement request, you should treat it as privileged information and respect its confidentiality, just as you would if you were asked to review it for a journal.

In practice, that's the standard you probably already know from peer reviewPeer reviewFormal evaluation of a paper by independent experts, usually organised by a journal or conference, before it is accepted for publication. arXiv moderation and endorsement are not peer review..

What confidentiality means in practice

Don't share it

• Don't forward the manuscript, or links to it, to colleagues, students or collaborators.
• Don't discuss its specific contents with others, even informally.
• If you'd like a colleague's opinion, ask the author first, and let them decide.

Don't use it

• Don't use the author's unpublished ideas, data, methods or text in your own work.
• If the paper overlaps with something you're doing, consider stepping back and saying so, rather than continuing to review.

Don't copy it into other services

• Don't upload or paste the manuscript into third-party tools, including AI chatbots or writing assistants. That shares the work beyond the people the author chose, and you often can't control how those services store or reuse what you give them.
• If you use tools to help draft your own feedback, keep the author's text out of them.

arXiv's policy on generative AI tools is about authors (it says authors are fully responsible for everything in their papers and should disclose significant AI use), but the same spirit applies to reviewers: you're responsible for what you do with someone else's work.

Don't keep it longer than you need

• Delete any local copies you downloaded once you've finished your feedback and decision.

Confidentiality continues after you decline

Whether you endorse, decline or step back, your obligation to keep the manuscript confidential stays the same. See How to decline kindly.

Endorsement decisions are private too

arXiv says whether you endorsed someone is private between you and arXiv's administrators and isn't shared with other users or the author, although an author might sometimes be able to infer it. Please respect that on your side too: don't publicise who you have or haven't endorsed.

How ResearchGuild protects manuscripts

We designed ResearchGuild so authors keep control of their work:

• Only people the author shares with can see a manuscript. It isn't public, and it isn't visible to other reviewers unless the author shares it with them.
• Manuscripts are deleted 30 days after the paper closes.
• Manuscripts are never used by Maximem for product, analytics or model training.

These protections only work if reviewers do their part. Downloading a copy, sharing it, or pasting it into another service takes it outside those protections.

If something goes wrong

If you accidentally share a manuscript or realise you have a conflict of interest:

• stop using or distributing it;
• tell the author honestly and promptly;
• delete any copies you control.

Owning a mistake quickly is far better than hoping nobody notices.

A simple test

Before doing anything with a manuscript, ask: "Would the author be comfortable if they saw me doing this?" If the answer is anything other than a clear yes, don't.

For the rest of your role, see What an endorser is (and isn't) responsible for and Giving useful pre-submission feedback. You can manage your account in settings.