Veridi
A structured methodology for checking facts
Veridi is a step-by-step fact-checking system built by a small Canadian organization. It provides explicit procedures for evaluating claims: classifying evidence by source quality, detecting disinformation techniques, and producing transparent, auditable assessments.
Tested against 97 carefully constructed claims. 96 passed. 1 partial. 0 failed.
The test suite included 24 adversarial scenarios engineered to exploit specific weaknesses, claims in four non-English languages, genuinely contested topics where experts disagree, and real-world disinformation patterns found in the wild.
The single partial result was a verdict that validated as correct with low confidence because a source was unavailable at test time.
We will continue testing and evaluating against real-world input, recognizing that a test suite cannot fully capture reality.
Where would you like to start?
I want to check a claim - Plain-language guide to what Veridi does and how to use it.
I’m a journalist or fact-checker - Source hierarchy, verdict taxonomy, and how Veridi fits into professional workflows.
Show me the data - Validation methodology, adversarial testing results, confidence calibration framework, and gaming countermeasure documentation.
I represent an organization - Mission, governance, impact potential, and partnership opportunities.
What makes Veridi different
Sources are ranked, not trusted. A four-tier hierarchy classifies every piece of evidence. Multiple low-quality sources cannot substitute for one high-quality source; structural confidence caps enforce this.
Gaming is expected. Eleven documented disinformation attack patterns have explicit detection procedures. The methodology treats fact-checking as an evolving adversarial problem, not a lookup task.
Institutions can degrade (or improve). An Institutional Reliability Index tracks when formerly authoritative sources have been compromised by political interference or defunding. A source that was Tier 1 last year may be Tier 3 today.
Everything is auditable. The full methodology is open for inspection. Every assessment shows its evidence, source tiers, confidence reasoning, and any gaming flags detected.
Open methodology
Veridi’s complete methodology files - including source hierarchy, decision trees, specialist frameworks, gaming countermeasures, and the Institutional Reliability Index - are available for review.