Example Assessments
These are real outputs from Veridi’s validation testing; not hand-picked successes, but representative examples showing how the methodology handles different types of claims.
Example 1: A confirmed true claim
Claim submitted: “More than 97% of actively publishing climate scientists agree that human activities are causing global warming.”
Veridi’s assessment:
VERDICT: TRUE — Confidence: 95%
Multiple independent meta-analyses confirm this figure. Cook et al. (2013) examined 11,944 peer-reviewed climate science abstracts and found that 97.1% of papers expressing a position endorsed the consensus that humans are causing global warming. Lynas et al. (2021) surveyed over 88,000 climate-related studies published between 2012 and 2020 and found greater than 99.9% consensus in the literature. NASA’s scientific consensus page cites both studies and lists eighteen scientific organizations - including the American Association for the Advancement of Science, the American Meteorological Society, and the National Academy of Sciences - that have issued statements affirming the consensus.
Sources: NASA Scientific Consensus page (Tier 1), Cook et al. 2013 in Environmental Research Letters (Tier 1), Lynas et al. 2021 in Environmental Research Letters (Tier 1), statements from eighteen scientific organizations (Tier 1).
Gaming flags: None detected.
Why this example matters: This is a straightforward case where the claim is well-supported by multiple independent Tier 1 sources. The confidence is 95% - the structural ceiling for multiple Tier 1 sources in agreement - because the evidence base is as strong as it gets: independent meta-analyses using different methodologies arriving at consistent results, confirmed by institutional consensus statements.
Example 2: A straightforward false claim
Claim submitted: “Exposure to 5G cellular towers causes or spreads COVID-19.”
Veridi’s assessment:
VERDICT: FALSE — Confidence: 95%
COVID-19 is caused by the SARS-CoV-2 virus and is transmitted person-to-person through respiratory droplets and aerosols. Radio waves - including those used by 5G networks - cannot transmit viruses. This is established physics: electromagnetic radiation at 5G frequencies (sub-6 GHz and mmWave) does not interact with biological pathogens in a way that could cause infection or transmission.
COVID-19 spread extensively in countries that had not deployed 5G networks, further disconfirming any causal link.
Sources: WHO (Tier 1), CDC (Tier 1), multiple peer-reviewed studies on radiofrequency exposure (Tier 1), PolitiFact investigation (Tier 2).
Gaming flags: None detected.
Why this example matters: This is a clean case: a false claim with overwhelming evidence against it. Veridi handles it efficiently and with high confidence because multiple Tier 1 sources agree. Not every claim is this straightforward.
Example 3: True facts assembled to mislead
Claim submitted: “India’s Home Minister Amit Shah said the share of Muslims in India’s population has been rising, pointing to Census data from 1951 to 2011 showing a steady decline in the Hindu share and a rise in the Muslim share.”
Veridi’s assessment:
VERDICT: MISLEADING — Confidence: 85%
India’s Census data does show that the Muslim population share rose from 9.8% in 1951 to 14.2% in 2011, while the Hindu share declined from 84.1% to 79.8%. These numbers are accurate.
However, the framing omits critical context that reverses the reasonable interpretation: Muslim population growth rates have been declining faster than Hindu rates. Between 1991 and 2011, the Muslim decadal growth rate fell from 29.4% to 24.7%, while the Hindu rate fell from 19.9% to 16.7%. The trends are converging, not diverging. Additionally, Shah’s attribution of growth to “infiltration” is unsupported: an RTI query confirmed the Home Ministry maintains no data on infiltrators identified or deported.
This is assessed as MISLEADING rather than LACKS CONTEXT because indicators of purposeful deception are present: cherry-picked metric (share vs. declining rate), advocacy context (political speech titled “Infiltration, Demographic Change, and Democracy”), and unsupported causal attribution.
Sources: India Census data (Tier 1), Pew Research Center demographic analysis (Tier 1), RTI response from Ministry of Home Affairs (Tier 1), Alt News investigation (Tier 2).
Why this example matters: Every fact in this claim is technically true. The Census numbers check out. But the facts are arranged to create a false impression - that Muslim population growth is an accelerating threat - when the data shows the opposite trend. Veridi distinguishes between “missing context” (an accidental omission) and “misleading” (a deliberate framing), based on whether the false impression appears to be the purpose of the claim.
Example 4: A genuinely unclear situation
Claim submitted: “The COVID-19 lab leak theory has been confirmed by U.S. intelligence agencies.”
Veridi’s assessment:
VERDICT: MOSTLY FALSE — Confidence: 85%
No U.S. intelligence agency has “confirmed” the lab leak theory. The word “confirmed” significantly overstates what any agency concluded.
In February 2023, the US Department of Energy assessed with “low confidence” that a lab leak was the most likely origin. The FBI assessed with “moderate confidence.” Four other agencies and the National Intelligence Council assessed that natural spillover from animals was most likely. The intelligence community remains split, and the assessments use explicitly hedged language: “low confidence” and “moderate confidence” are technical terms indicating significant uncertainty.
The claim contains a true element: some agencies did assess that a lab leak was more likely than not. But the core assertion - that the theory has been “confirmed” - is false. Confirmation is categorically different from a low-confidence assessment.
Sources: Office of the Director of National Intelligence declassified assessment (Tier 1), Congressional Research Service summary (Tier 1), PolitiFact analysis (Tier 2).
Why this example matters: This claim isn’t simply true or false. Some agencies did lean toward the lab leak hypothesis. But the word “confirmed” transforms an uncertain assessment into a definitive conclusion that doesn’t exist. The MOSTLY FALSE verdict acknowledges the true elements while identifying that the core claim is inaccurate.
Example 5: Detecting disinformation techniques
Claim submitted: “The World Health Organization’s International Agency for Research on Cancer classified processed meat as a Group 1 carcinogen in 2015 - the same category as tobacco smoking and asbestos exposure - meaning that eating a daily serving of bacon or sausage carries cancer risks comparable to a pack-a-day smoking habit.”
Veridi’s assessment:
VERDICT: MISLEADING — Confidence: 92%
The IARC did classify processed meat as a Group 1 carcinogen in 2015. Processed meat is in the same IARC group as tobacco smoking and asbestos. These facts are accurate.
However, the conclusion - that the cancer risks are “comparable” - is false, and the false impression is created by conflating two different things: evidence strength and risk magnitude. IARC Group 1 means “there is convincing evidence that the agent causes cancer.” It does not mean “the agent causes cancer at a similar rate to everything else in the group.” The WHO itself clarified this distinction at the time of the classification.
The actual risk difference is enormous. Eating 50 grams of processed meat daily increases colorectal cancer risk by approximately 18%. Smoking a pack of cigarettes a day increases lung cancer risk by approximately 2,000%. The risks are not comparable; they differ by roughly a factor of 100.
Gaming flags detected:
- Framing manipulation: True facts about the IARC classification are assembled to create a false impression of risk equivalence. The manipulation works by conflating evidence strength (how sure we are it causes cancer) with risk magnitude (how much cancer it causes). Each individual statement is accurate; the composite is false.
- Anchoring: The true, easily verified facts (Group 1 classification, same category as tobacco) serve as anchors that transfer credibility to the false conclusion (comparable risk levels).
Sources: IARC Monographs Volume 114 (Tier 1), WHO Q&A on the classification (Tier 1), Global Burden of Disease Project estimates (Tier 1).
Why this example matters: This is a stress test: every individual fact in the claim is true. The IARC classification is real. The group assignment is real. The comparison to tobacco’s group is real. Nothing is fabricated. But the facts are arranged to create a false impression about risk magnitude by exploiting the gap between how scientists use the word “carcinogen” and how the public understands it. Veridi detected the framing manipulation and anchoring, quantified the actual risk difference, and cited the WHO’s own clarification as primary evidence.