Google Fact Check Scraper

Search Google's Fact Check Tools database — the same index that powers Google's fact-check labels — and export every reviewed claim, the publisher, the rating, and the review URL to a clean dataset.

What you get

• One dataset row per claim review with claim text, claimant, claim date, publisher name, publisher site, review title, review URL, review date, textual rating, and language
• Multi-query — pass a list of topics, people, or claim phrasings and get one consolidated dataset
• Filter by publisher — restrict to a single fact-checker site (Snopes, PolitiFact, FactCheck.org, AFP, Reuters Fact Check, etc.)
• Filter by recency — limit to claims reviewed within the last N days
• Multilingual — restrict by language code or pull every language
• Auto-pagination — the actor walks the API's nextPageToken chain until your result cap is hit

Use cases

• Misinformation research — pull every fact-checked claim about a topic, candidate, or event
• Journalism prep — surface what fact-checkers have already covered before publishing
• AI trust & safety — build a labelled dataset of claims with verdicts for model evaluation or RAG grounding
• Brand and reputation monitoring — track fact-checks mentioning a company or product
• Election integrity — assemble a feed of newly reviewed political claims by date

How to use

1. Add your topics, names, or claim phrasings to Queries — one per line.
2. (Optional) Restrict by Language code (e.g. en, es, pt).
3. (Optional) Filter to a single fact-checker via Publisher site filter (e.g. snopes.com, politifact.com).
4. (Optional) Limit by Max age (days) to focus on recent reviews.
5. Set Max results per query (default 50, max 500). The actor paginates 10 at a time.
6. Run — each reviewed claim becomes its own dataset row.

Output schema

Field	Type	Description
query	string	The input query
rank	number	1-based rank within the query (0 if no results/error)
status	string	success, no-results, or error
claimText	string | null	The claim being fact-checked
claimant	string | null	Who said it (politician, public figure, social-media account)
claimDate	string | null	When the claim was made (ISO timestamp)
reviewPublisherName	string | null	Fact-checker name
reviewPublisherSite	string | null	Fact-checker domain
reviewUrl	string | null	URL of the published fact-check
reviewTitle	string | null	Title of the fact-check article
reviewDate	string | null	When the fact-check was published (ISO timestamp)
textualRating	string | null	The verdict in the publisher's own words ("False", "Pants on Fire", "Mostly True", …)
languageCode	string | null	Language of the review
error	string | null	Error message if status is error

If a claim has multiple reviews (e.g. reviewed by both Snopes and PolitiFact), each review is its own row — same claim text, different publisher/rating/URL.

Tips

• Phrase queries broadly — search "vaccine" rather than a specific debunked claim; the API matches across claim text and review text.
• Ratings are not normalised — textualRating is whatever the publisher wrote. Map them to your own scale downstream.
• Coverage skews English and US-centric, but the API supports many languages — try Spanish or Portuguese for Latin-American coverage.
• Use Publisher site filter to build a single-publisher feed for a daily digest.
• Some claims have a claimDate years before the reviewDate — useful to spot resurfaced misinformation.