For researchers

Check your references before a reviewer does

AI writing tools invent citations — plausible authors, real-looking DOIs, papers that were never written. arXiv suspends authors over it. Paste your bibliography into the health check and every reference is checked in your browser: does it resolve, does the title match its DOI, is it marked retracted. Free, unlimited, no account — and your bibliography is never uploaded; no document ever reaches us.

Why in-browser matters: the checkers you have heard of ask you to upload your reference list to their servers. Ours parses in your browser and looks each reference up one identifier at a time, straight from your machine to the open registries (Crossref, DataCite, OpenAlex) — open your network tab and watch. There is no server-side copy of your bibliography to retain, leak, or train on. And because the check covers DataCite as well as Crossref and OpenAlex, references to Zenodo, Dryad, and figshare deposits resolve here instead of getting flagged as missing.

Then put your work on the record

A public page for your paper — every identifier in the graph gets one, listing what your work cites and what cites it, each claim labelled signed or inferred.
Citations that say something. Plain “cites” is flat. Typed claims record that a paper supports yours, refutes a result, extends a method, uses your dataset.
A priority trail. Signed claims are timestamped and chained on a public log. When it matters who said what first, the record is already public and already ordered.
A place AI assistants check. CiteStamp runs a live endpoint AI agents query before they cite — so when an AI summarizes your field, there is a public, labelled record to consult instead of a guess.

A public paper page: the claim button on top, citation claims labelled signed or inferred listed below — your paper’s public page — the claim button sits right on top

Claim your paper in two minutes

Machine sweeps of Zenodo and SSRN metadata have already mapped citations for hundreds of thousands of works. Search your DOI on the main page, open the paper page, and follow Claim with ORCID — pick the citations that are really yours and put your name on them. Your ORCID sign-in proves identity; authorship is checked against the work’s own metadata, and a human curator handles anything the metadata can’t settle.

The claim page for a work already in the graph: the count of machine-inferred citation claims on it, and the Sign in with ORCID button that starts the claim — two minutes, start to signed

What it costs

The checker, the graph, the paper pages, and signing your own work: free. Pro is $35 a year — a higher-rate API token for people and programs that hit the graph hard, and a way to keep an independent citation graph running. The pricing page lists exactly what Pro includes today; nothing on it is a preview. Signing keys for the full minting surface are issued by hand while the service is small: info@citestamp.com.

What we will never do with your data

There is almost none to misuse. The graph stores links between public identifiers — no full text, no abstracts, no personal data on the record. Claims are CC0 and the whole log is publicly forkable, so the record outlives any one operator, including us.

Start here: check a bibliography — or ask the graph about your DOI; if your paper is in the sweep, the claim button is one click away.