Before You Forward It,
Verify It.
Verify suspicious messages, screenshots, voice notes, PDFs, and links with evidence-backed explanations in your language.
How SatyaSetu Verifies Claims
A four-stage deterministic workflow powered by authoritative source grounding and atomic claim extraction.
1. Submit Content
Forward WhatsApp messages, upload circular PDFs, audio clips, or paste news links into our portal.
2. Extract Claims
Multi-claim statements are decomposed into atomic, verifiable propositions to isolate falsehoods from truth.
3. Retrieve Evidence
Queries Tier-1 Gazettes, PIB Fact Check, and statutory archives. Verified against Shared Rumour Memory.
4. Deliver Verdict
Receive an authoritative verdict, verbatim quotations, and direct links to official sources in your language.
The Five Evidentiary Verdicts
SatyaSetu adheres strictly to IFCN fact-checking criteria. We never guess: when evidence is insufficient, claims are marked as unconfirmed.
Statutorily Accurate
Supported by active official gazette notifications, primary circulars, or certified government announcements.
Factually Refuted
Directly contradicted by official public records, statutory notices, or confirmed as fabricated disinformation.
Expired Context
May have been true in a previous financial year or pandemic period, but circulated today as current truth.
Mixed Evidence
Contains a kernel of truth (e.g., scheme exists) mixed with inflated grant amounts or phishing registration links.
Insufficient Evidence
When official records are absent or private claims lack public evidence, SatyaSetu abstains from speculation.
Engineered for Real-World Civic Misinformation
Every feature below is implemented in code and operational across our web portal and WhatsApp bot.
Multimodal Input Ingestion
Accepts text forwards, screenshot images with OCR, audio voice memos, PDF circulars, and article URLs.
Atomic Claim-Level Correction
Decomposes mixed forwards into discrete statements so legitimate facts aren't mistakenly dismissed alongside false claims.
Evidence-Grounded Verdicts
Every verdict cites Tier-1 gazettes, official .gov.in domains, and verified IFCN repositories with verbatim quotes.
Shared Rumour Memory
Identifies viral forwards using L1 caching and L2 vector memory, returning instant verified responses to curb spread.
Multilingual Reasoning
Supports verification and explanation in Hindi, Marathi, and English with cross-lingual claim mapping.
Twilio WhatsApp Integration
Direct tipline integration via official webhook with verified OTP delivery and user session management.
Civic Impact & Digital Literacy Goals
Our mission is to empower citizens with evidentiary truth before harmful forwards are amplified.
Language Inclusion
Ensuring citizens accessing regional languages receive equal evidentiary protection from viral deceptions.
Media Literacy
Training readers to spot phishing URLs, forged government circulars, and out-of-context video claims.
Financial Scam Prevention
Intercepting fake welfare recruitment scams and fraudulent lottery schemes before citizens suffer losses.
Rapid Fact-Check Triage
Substantially reducing turnaround time from days to seconds with deterministic rules and authoritative search.