PRUUF instantly rates any news article across 8 veracity dimensions — factual accuracy, bias, manipulation, source quality, and more — powered by AI and grounded in real-time research.
No setup, no subscription. Install the extension and PRUUF works whenever you need it — on any article, any site.
Browse news as you normally would. PRUUF works on any news page you visit — no configuration needed.
Tap the PRUUF icon in your browser toolbar. Our AI engine analyzes the article across 8 veracity dimensions with real-time web grounding.
Get an instant 0–100 score, a plain-language verdict, and a breakdown of exactly where the article excels or falls short.
PRUUF analyzes each article across eight independently scored dimensions — a complete picture, not just a headline number.
Are verifiable claims correct, current, and supported by reliable evidence? Errors, outdated facts, and fabrications are penalized.
Does the article use loaded language, straw-man framing, false urgency, or other rhetorical devices to mislead?
Does the article's structure suggest intent to inform or deceive? Sensationalism and propaganda patterns are flagged.
Are sources credible, traceable, and clearly cited? Anonymous sourcing and vague references reduce the score.
Does the article leave out information a reasonable reader needs? Cherry-picking and selective framing are penalized.
Is the reasoning sound? Logical fallacies, unsupported leaps, and conclusions disproportionate to evidence are caught.
Is the article logically self-consistent? Contradictions and conflicting claims within the same piece are penalized.
Does the article clearly separate facts from interpretation? Opinion presented as fact is flagged — opinion itself is not.
PRUUF doesn't guess. Every analysis is verified through a four-layer grounding system before you see a result.
All numeric scores are independently recalculated and cross-checked before the final result is computed. No arithmetic errors slip through.
Key claims are researched against live web sources before scoring. The AI cites what it found — and penalizes claims it couldn't verify.
Facts from previous analyses are stored as vector embeddings. New articles are checked against this library for consistency.
Low-confidence results are flagged for expert review. Manual overrides from our editorial team are incorporated into future scoring.
A bar that stays out of the way, a panel when you want the verdict, and the full scorecard when you want to argue with it. Drag it wherever you like.
Free to install. Works on Chrome, Edge, Brave and Firefox. No account required.