sourcerated.com

How We Verify

 

 

 

1. How We Verify

SLUG: how-we-verify

▸ Pages → Add New  |  Title: “How We Verify”  |  Slug: how-we-verify  |  Paste content below using the Text/HTML editor tab

PLATFORM

How SourceRated Verifies Every Story

SourceRated uses a four-step AI verification pipeline that runs automatically on every article published to this platform. No story goes live without passing through this process.

Step 1 — Source Detection

Our AI crawlers monitor RSS feeds from 24 globally recognised news outlets every 20 minutes. When a new story is detected, it is immediately logged with its source, headline, publication timestamp, and content snippet.

Step 2 — Cross-Reference Check

The AI searches all other monitored sources for the same story. It identifies matching stories based on headline similarity, named entities (people, places, organisations), and event descriptions. The number of independent sources reporting the same event is the primary input into the credibility score.

Step 3 — Credibility Scoring

Based on the source cross-reference, each article is assigned a credibility score from 0 to 100. The scoring table below shows exactly how scores are calculated:

Sources corroborating Score range Label
5 or more independent sources 90–100% Very High Credibility
3–4 independent sources 70–89% High Credibility
2 independent sources 50–69% Moderate
1 source only 20–49% Unverified
Contradicted by other sources 0–19% Disputed

Step 4 — Community Verification

Every article includes a community voting widget where readers can vote TRUE, UNSURE, or FALSE. This crowd-sourced signal is displayed alongside the AI score as a second independent trust indicator. The two signals together — AI credibility score and community verdict — give readers the most complete picture of a story’s reliability available on any news platform.

SourceRated never removes articles for having a low credibility score. Low-scoring stories are clearly labelled and displayed with warnings. Transparency is always preferable to suppression.

What We Do Not Do

  • We do not editorially decide which stories are true or false
  • We do not remove stories because they are controversial
  • We do not accept payment to influence credibility scores
  • We do not reproduce full copyrighted articles — all content is AI-summarised from public sources