The Fact Checker Built for Video

When you read an article, you can skim for claims, copy a sentence into a search engine, and check a source in under a minute. Video doesn't work that way. Misleading claims in a YouTube video or Vimeo documentary are embedded in speech, layered with emotional footage, and consumed passively. Most viewers can't — and don't — pause to verify every statistic or assertion made by a presenter. The result is that false or misleading information in video travels further and faster than in any other format. A 2023 Reuters Institute study found that video was the format most associated with misinformation encounters across all age groups. Truthify exists to close that gap.

Truthify does not analyse visuals, audio quality, or presenter tone. It analyses the spoken claims — the factual assertions made in the video's dialogue. This includes: statistics quoted by the presenter, historical claims and dates, scientific statements, geographical and demographic data, quotes attributed to other people or organisations, and comparative claims ("X is bigger than Y", "Z is the largest/first/only"). Opinions, predictions, and value judgments are identified and labelled as opinions rather than being fact-checked, since they cannot be verified against evidence.

After processing a video, Truthify calculates two summary metrics. The Trust Score is a percentage showing what proportion of verifiable claims are supported versus refuted or misleading — higher is better. The Confidence rating (Very Low to Very High) reflects how decisively the evidence speaks for each verdict — a claim with ten corroborating academic sources has higher confidence than a claim where the only evidence is a single news article. These metrics give you a quick signal about whether a video is broadly reliable without reading every individual card.