What we learn from watching AI engines describe, recommend, and cite entities — published from the Entidex platform.
People are told to cross-check answers across AI engines and trust the consensus. We measured the consensus: across every fact two or more engines both answered about real tracked entities, how often their answers conflict — per-engine-pair disagreement rates with explicit sample sizes and 95% intervals, plus why engine consensus is not the same thing as truth.
A live per-engine leaderboard of record-agreement — how often each engine’s answers about real tracked entities match the verified record. Pooled figures with explicit sample sizes and 95% intervals; engines with thin coverage are held, never ranked. Re-aggregated daily from stored evaluations.
We measured 50 creators and shows against the verified record — 503 checkable facts. Mean record-agreement was 71.7%, 76% of the cohort carried at least one wrong fact, and the per-engine spread is real (Grok 78.9% vs ChatGPT 71% on the full cohort; sample sizes stated for the rest). Method, findings and honest caveats.
We caught an AI engine laundering our verified entity data — accurate facts, fabricated provenance. Three weeks later a German court described the same mechanism in Google’s AI Overviews and called it the operator’s own liability.
Run a free scan — no signup, no key. Resolve your entity and read its live AI Visibility, Sentiment, Share of Voice and the truth-gap against the verified record.