Study ties AI answer trust directly to the news brands it cites
What this answers
- Do people trust AI answers more when they cite trusted news brands?
- Yes — an AOP/Ipsos survey of 1,000 UK adults found trust in an AI answer averaged over 90% when it cited a fully trusted source, versus about 10% for a distrusted source and 25% for a neutral one.
- Do people realize AI tools can fabricate sources?
- Not always — 37% of respondents didn't realize AI tools can fabricate information or sources, rising to 45% among those aged 45-54, per the same survey.
The Association of Online Publishers’ AI Publisher Impact Study, conducted by Ipsos among 1,000 UK adults and reported by Press Gazette, finds trust in an AI-generated answer tracks the trustworthiness of the source it cites: over 90% when the source is fully trusted, about 25% when neutral, and roughly 10% when distrusted. Distrusted citations also produced far fewer click-throughs, while highly trusted ones sometimes led users to stop at the AI answer without clicking further. Separately, 37% of respondents said they didn’t realize AI tools can fabricate information or sources, rising to 45% among those aged 45-54.
Why it matters: Publisher brand trust is turning out to be a direct input to AI answer credibility, raising the stakes of which sources engines choose to cite.
Posted to the wire August 1, 2026. Edited by Joe Balewski.