TRUTH VERIFICATION ENGINE

The heart and soul of Candor.

Shield verifies every piece of information the system ingests — evaluating, weighing, and rendering a percentage of truthfulness. Truth scores accumulate on users, driving algorithmic promotion. The most honest voices get the most reach.

Part of the Candor Safety Ecosystem
CAPABILITIES
What Shield does.
Every capability is designed to work autonomously — detecting, responding, and protecting without human intervention.
⚖️
Content Truth Scoring
Every post, document, and media upload is analyzed by AI. Each piece receives a truthfulness percentage that is transparent, auditable, and attached to both the content and the user.
📈
User Reputation Engine
Truth scores accumulate over time. Users who consistently post verified content build reputation. That reputation drives algorithmic promotion in Reach. Truth is rewarded structurally.
🔍
Classified Document Detection
Shield scans all ingested documents for government classification markings — TOP SECRET, SECRET, CONFIDENTIAL, NOFORN, SCI. Flagged documents are held, never published without evaluation.
🤖
AI Escalation Chain
Flagged content escalates through three tiers: Claude Haiku (fast first pass) → Claude Sonnet 4.6 (deep analysis) → human review. Each tier applies more rigorous evaluation.
🏦
Secure Vault
Sensitive flagged documents are quarantined in a secure data area. Time-sensitive material is automatically destroyed after 24 hours. No classified content sits on servers indefinitely.
📰
Press Freedom Infrastructure
Journalist and whistleblower subscriptions provide anonymous publishing. The bedrock of freedom of the press — publish without fear of identification.
"Shield doesn't decide what's true. It measures truthfulness and puts the score in your hands. You decide."
HOW IT WORKS
The pipeline.
From detection to response — fully automated.
Content Ingested
Every piece of content — text, images, documents, video — enters the Shield analysis pipeline.
AI Analysis
Claude Haiku performs rapid first-pass evaluation. Truthfulness score generated with confidence metrics.
Classification Check
Documents are scanned for government classification markings. Matches are quarantined and escalated.
Score Published
The truthfulness score is attached to the content and accumulated on the user's reputation. Transparent and auditable.
COMPLIANCE & STANDARDS
Built for the real world.
NCMEC ESP
CyberTipline Approved
Registered Electronic Service Provider. CSAM detected by Shield triggers mandatory federal reporting under 18 U.S.C. § 2258A.
AUDITABLE
IPFS Audit Trail
Every Shield verdict is stored on IPFS and recorded on the Polygon blockchain. Anyone can verify any decision. Forever.
24HR DESTROY
Secure Vault Protocol
Time-sensitive classified documents are automatically destroyed after 24 hours. Zero long-term storage of sensitive material.
Ready to integrate?
Contact us for API access, enterprise licensing, or grant partnership opportunities.
[email protected]

Choose Your Plan

Content truth scoring for newsrooms and publishers

INDIE
$200
/month · 50K scans · 99.9% SLA
NEWSROOM
$1,000
/month · 500K scans · 99.99% SLA