Paste any text. We extract the discrete factual claims and assess each one against Claude Sonnet 4's knowledge. Free, no signup.
Up to 500 words on the free tier, 2,000 signed in. Best for AI-generated drafts, research summaries, news excerpts, and reports.
Each factual claim is extracted, color-coded by assessment (verified, disputed, unverifiable, inaccurate), and explained in plain English.
Anything marked unverifiable or disputed is your manual check list. We tell you exactly what to verify and what kind of source you'll need.
Yes. 3 checks per day with no signup. Sign in (free account) for 20/day. Caps are tighter than the other free tools because each fact-check involves extracting claims and reasoning over each one — a heavier LLM call.
No — assessments are based on Claude Sonnet 4's training-data knowledge (cutoff: early 2026). For time-sensitive claims (recent events, current statistics, breaking news), the tool marks them as unverifiable and recommends manual verification with primary sources. We may add live-web grounding to a paid tier later.
A search engine returns links — you still have to read each one and decide whether the claim is supported. This tool extracts the discrete factual claims from your text upfront and assesses each one. Especially useful for AI-generated drafts where hallucinations look authoritative but reference invented studies, dates, or statistics.
Verified: matches well-established knowledge. Likely true: probably correct, not high confidence. Disputed: contested in mainstream sources. Unverifiable: needs live data or primary sources to confirm. Inaccurate: demonstrably wrong. Opinion: a value judgement, not a checkable fact.
500 words on the free tier. 2,000 words signed in. For longer documents, fact-check in sections (introduction, key claims paragraph, conclusion).
For well-established knowledge in domains the model is strong on (history, geography, mainstream science, well-documented events), assessments are reliable. For specialist domains, recent events, or rapidly evolving fields, treat the output as a triage tool — it tells you which claims need a closer look, not which ones are definitively true.
3 free scans/day, no card. See how AI-detectable your fact-checked draft is.