
AI systems use news data to create value, and in some cases the publisher gets nothing in return. Where there is a deal, the price cannot be explained. The market has an agreed way to meter usage, an agreed way to report it, yet no agreed way to value it.
This paper sets out a way to measure that value, and reports what the measurement found. The results come from 750 questions over 25 days in October 2026. These questions were posed to three AI models, evaluated by three AI judges, and used articles from more than 40 publishers in Ozone’s portfolio.
The comparison produced four findings: whether publisher content enhances AI responses at all, how that gain changes as more content is supplied, whose content matters, and how credit should be measured.
Sign up below to download the full research and explore the findings.
