Claim & Citation Checker
Paste any text, AI-written or not, and see the factual, numeric, and absolute claims that lack a source. A prompt to go verify them yourself. It can't tell you what's true; it tells you what to check.
What this does, and what it deliberately doesn’t
This tool does not detect hallucinations or judge whether anything is true, no client-side tool honestly can. What it does is surface the sentences that assert something checkable, statistics, superlatives, dates, named authorities, and flags the ones with no citation. Those are exactly the lines worth verifying before you publish, whether a human or an AI wrote them. Think of it as a highlighter for “go check this,” not a truth machine.
What gets flagged
- Numbers and stats, percentages, counts, dates, “millions”.
- Absolutes and superlatives, “the best”, “always”, “everyone”, “proven”.
- Vague attributions, “studies show”, “experts say”, with no named, linkable source.
A sentence with a real citation (a link, “according to [named source]”, a footnote) is marked as sourced, not because it’s correct, but because there’s something to check. The judgment is still yours. Nothing is uploaded.
FAQ
Does it tell me if a claim is false?
No, and it never claims to. It can’t know truth. It points at the statements that make a checkable assertion without backing them up, so you know where to look.
Why not call it a hallucination detector?
Because a heuristic can’t detect hallucinations, and a tool that pretended to would mislead exactly the people who’d trust it. Honest scope: flag unsupported claims, leave the verifying to you.
Related
Tighten AI output with structured outputs, or understand why models assert confidently in how context works.