What is Factual Inaccuracy Rate, and how is it different from hallucination?

A factual inaccuracy is a mention that references you accurately but carries materially wrong information — stale pricing, a superseded spec, a misattributed feature. The model is faithfully reflecting a source; the source is out of date or incorrect.

This is the distinction that decides the remedy. A hallucination is an AI platform quality problem you cannot fix directly. A factual inaccuracy traces to content you own, which means it is fixable, and the IAB is explicit that the source of truth must be brand-supplied.

That is the loop Senso is built for: the Context Layer is the brand-supplied source of truth, evaluation finds the stale fact or contradicted claim, remediation corrects the content, and publication puts the corrected version where the models read it.