Methodology
Every Crosscheck fact-check names the exact claim it evaluates, assigns one of seven verdicts, and cites the primary or independent sources behind that rating. Verdicts reflect the public record at the time of publication and are revised when the record changes.
What we check
We evaluate specific, checkable factual claims made in public political discourse — statements that can be confirmed or refuted against evidence. We do not rate opinions, predictions, or value judgments. When a single statement bundles several claims, we separate and rate each one, which is why a single article often carries a mix of verdicts.
How we rate
Each claim receives one rating on this seven-point scale. The full verdict index shows every fact-check grouped by rating.
- True Accurate and supported by the available evidence.
- Mostly True Accurate in substance, with minor caveats or missing context.
- Mixed Contains both accurate and inaccurate elements — the truth is partial.
- Unproven No conclusive evidence either way; the claim can't yet be verified.
- Misleading Individually defensible facts arranged to leave a false impression.
- Contested Genuinely disputed by credible sources; reasonable people disagree.
- False Contradicted by the available evidence.
Sourcing
We prefer primary documents — court filings, official records, transcripts, datasets — and independent reporting over secondary characterization. A claim is rated only when its sources are reachable and actually support the rating assigned. A source that blocks automated access is verified through an alternate retrieval, not dropped. Every cited source is linked so readers can check our work directly.
Corrections & updates
Errors are corrected in the open. We distinguish a correction (we got something wrong and fixed it) from an update (the record changed and we added new information) — and never relabel one as the other. Each is logged with a dated note on the article itself and listed on the corrections page.
If you think a verdict is wrong, write to the desk with the claim and the evidence you think we missed. We re-check it against sources. If the challenge holds up, the rating is revised and the article carries a dated note saying what changed and why. Subjects of a fact-check get the same route as any other reader, and no more.
How this publication is produced
Crosscheck uses AI systems to do the labor of fact-checking — monitoring public claims, retrieving sources, and drafting articles. No verdict is published without human review. Every draft goes to a human editor who checks the sources, the reasoning, and the rating before approving publication; drafts that don't hold up are revised or rejected. AI does the research; a person signs off on every rating. Errors that survive that review are corrected openly under the policy above.
Archive & republication
Crosscheck moved to this domain in June 2026. Articles dated earlier than that were first published by the same editorial project at its previous address, and were republished here with their original publication dates intact. Those dates are when the work first appeared, not when it was moved, and the originals remain online. Every republished article records where it came from. We will point any editor, researcher, or subject of a fact-check to the original on request.
Who we are
Crosscheck is written and edited by The Crosscheck Desk, an independent editorial project. It is operated by a single individual based in the United States, and is not incorporated. One person selects the claims, reviews every draft, and holds final say on every verdict. No outside party reviews, approves, or has any right of approval over what is published here.
We publish pseudonymously, and the reason is personal safety: fact-checkers working on political claims are routinely targeted for harassment. A reader does not need the author's identity in order to check the work. A reader needs the claim, the rating, and the sources, and all three are on every page. Judge us on whether the cited sources actually support the verdict.
Reach the desk at [email protected]. Mail is read by the same person who edits the verdicts.
Ownership & funding
Crosscheck is self-funded. Its running costs are web hosting and the AI services used for research and drafting, and they are paid personally by the operator. The publication earns no revenue.
We take no money from anyone else. Specifically: no advertising, no sponsored posts, no paid placement, no affiliate links, no donations, no grants, no paywall or paid subscription, and no payment of any kind in exchange for coverage or for a rating. We are not affiliated with any political party, campaign, candidate, PAC, or advocacy organization, and no such group has any influence over what we check or how we rate it.
If any of that changes, this section changes with it and the change is dated on the corrections page.
Privacy & analytics
We use Cloudflare Web Analytics, which counts page views without cookies and without fingerprinting individual visitors. We do not run Google Analytics, advertising trackers, or any third-party ad network, and there is no advertising on the site to track you for.
The email digest is free. Your address is used only to send it, is never sold or shared, and every issue carries a one-click unsubscribe.
This is the summary. What we collect, how long we keep it, and how identifiers are hashed are set out in full on the privacy page.
Submit a claim
Spotted a claim worth checking? Send it to us. We review every submission, though we can't promise to cover them all.