VeriFact® is SVN's end-to-end verification engine. Every submission moves through six evaluation layers — answering the Who, What, When & Where, Why, and How — before a senior editor signs off and the metadata record is preserved on-chain.
VeriFact begins by evaluating where the video originated — including uploader account history, prior activity, and available signals of identity or credibility. Verified journalists and consistent publishing behavior increase confidence, while anonymous or newly created accounts are reviewed more cautiously.
The video is compared against independent sources wherever possible, including additional footage, eyewitness accounts, and reporting from trusted outlets. Consistency across multiple independent sources strengthens confidence; contradiction triggers deeper review.
Authenticity depends not only on what is shown, but when and where it was captured. VeriFact cross-references timestamps, geolocation clues, weather patterns, and known events to determine whether the video aligns with real-world conditions at the claimed moment.
Advanced models identify potential synthetic or AI-generated elements — including deepfakes, voice cloning, and visual anomalies. This step helps determine why the video was created: whether it is legitimate news, misinformation, propaganda, political manipulation, or a scam.
VeriFact examines the video file itself for signs of tampering — frame-by-frame analysis, compression artifacts, and inconsistencies in visual or audio patterns. Signals suggesting edits, splicing, or re-encoding are flagged for deeper review.
Trained editors review the findings from each layer — assessing nuance, context, edge cases, and anything automated systems may miss. Human judgment ensures each published verification reflects both technical analysis and editorial standards. No video reaches publication without a senior editor's sign-off.
Journalists upload video through a dedicated submission page. They receive email notification when a decision has been reached and the video has been scored.
Once submitted, the video automatically runs through the verification system, applying source-authentication, cross-source corroboration, context and chronology checks, AI and manipulation detection, and content integrity analysis before any human review begins.
The video receives a reliability confidence score on a 0–100 framework. Scoring categories include:
An editor reviews the automated findings, checks the criteria, investigates questionable facts, adjusts scores where appropriate, completes the metadata verification, and sends the final video into the content pipeline. After Managing Editor review, the video can be published to SocialVerifiedNews.com and SVN social platforms, with the metadata verification page entered onto the blockchain as the permanent audit record.
After a video clears Human Editorial Review and is approved by the Managing Editor, its metadata verification page — the audit trail of every layer, score, signal, and editorial note — is written to the blockchain. The chain's job is preservation, not adjudication.
The blockchain does not verify truth. VeriFact verifies. What the chain provides is an immutable, timestamped record of how the verification was reached — so that months or years later, any reader, regulator, or researcher can audit the same evidence the editors saw.
See an example record: VeriFact Block Scoring →