From headline to Proof Score in minutes. We extract claims, find corroboration, analyze tone, and publish transparent reports—so readers and newsrooms can verify before they amplify.
Five stages. Built for speed, explainability, and accuracy.
Paste a link, upload a PDF, or drop raw text. We extract main content, remove boilerplate, and detect the language and structure.
NLP identifies verifiable statements, entities, numbers, and quotes, mapping each claim to its supporting sentences.
We retrieve diverse sources and run stance analysis to see if each claim is supported, contradicted, or uncertain.
We analyze tone, framing, hedging, and sentiment to highlight where language departs from evidence.
Per‑claim assessments roll up into an article‑level Proof Score with an explanation and export options.
*Advanced OCR and media forensics via SignalGuard™ in higher tiers.
An explainable pipeline—modular, auditable, and fast.
Run multiple models and data sources concurrently; compare stances; compute consensus‑weighted scores.
Suggests stronger or more current sources and reduces echo‑chamber effects.
Inline overlays for tone, sentiment, and framing—without changing the underlying facts.
Isolated routing with organization‑owned keys, prompts, and policy guards.
Attach signed verification manifests readers can independently validate.
Export full reports and evidence bundles for archives and audits.
// Submit an article for verification
POST /v1/verify
Content-Type: application/json
{
"url": "https://news.example.com/article.html",
"options": { "biasLens": true, "export": ["pdf","json"], "model": "balanced" }
}
// Retrieve a report
GET /v1/reports/{report_id}
Webhooks and signed manifests available. Region‑locked processing for enterprise.
Works where you do—browsers, CMS, and code.
Highlight claims on-page, preview Proof Scores, and export to ProofStack™.
Pre‑publish checks with status badges; WordPress/Ghost hooks; StoryGuard™ webhooks.
ProofStack™ bundles (PDF/CSV/JSON) and signed manifests via Certify™.
Scores summarize evidence as of retrieval time—never an appeal to authority. Every score links to evidentiary context.
Facts evolve. StoryGuard™ watches for updates; TruthTrail™ tracks versions; Certify™ regenerates manifests on change.
Sample results (fictional data):
Claim contradicted by clinical summaries; headlines overstate evidence.
Partial support in one region; global numbers inconsistent or out‑of‑date.
Strong alignment with peer‑reviewed summaries and official data.
Try BulletPROOF on your next article and export a fully transparent report.
Get Started