Does trust in AI chatbots drive news-seeking behavior?

The Reuters Institute's Digital News Report 2026 reports that weekly use of standalone AI chatbots for news "rose from 7% to 10% globally since last year," with growth concentrated in "parts of Asia, Africa, and Latin America, as well as Southern and Eastern Europe." Only 1% name AI their main source of news, so the report calls this "a complementary role for most users." Against this backdrop of low general trust in news (37% trust most news most of the time), trust in AI chatbots for news sits lower still at 20% globally — but the report finds a sharp split between users and non-users: "44% of AI chatbot users express trust in news from AI chatbots, compared with just 17% of non-users." At the market level, the report finds "a strong relationship" between a market's trust in AI chatbots and its rate of chatbot use for news, and this relationship holds "considerably stronger" (higher R²) for AI chatbots than for social media; the same pattern holds for individuals, who are "several times more likely" to use AI chatbots for news if they trust them, a weaker effect for social media.

The report's own explanation is that chatbot use for news is "still an emerging and more deliberate behaviour," so a person's trust gates whether they opt in at all. Social media use for news, by contrast, is mostly incidental — folded into a platform people visit for many other reasons — so low trust does not stop exposure the way it stops a deliberate choice to ask a chatbot. The report is explicit that all of this is self-reported, carrying normal recall and social-desirability caveats, and that findings describe a correlation across people and markets, not a causal test of whether trust produces use or use (and any resulting bad experience) revises trust downward.

This gives Does chatbot language style actually shape how much we trust it? a population-scale number for the question it leaves open. That paper argues chatbots earn trust rhetorically, through "natural language signifiers of expertise," but states plainly that it has "no sample, no study design, and no measured effect." The Digital News Report cannot show that expertise signifiers are the mechanism, but it does show, across 45 markets, that trust and chatbot adoption for news move together more tightly than trust and social-media use do — the kind of population-level pattern a signifiers account would need to explain. It also sits beside Do AI writing tools improve online discussion or degrade it?, which found AI tools raising engagement while cutting perceived authenticity in an experimental forum setting; here the self-selection runs the other way, since people who already trust chatbots are the ones choosing to use them for news, rather than a tool changing trust mid-interaction.

The excerpt does not test whether chatbot use itself raises trust over time, whether the 44%-versus-17% gap reflects selection (trusting people adopt) or experience (adopting raises trust), or what specifically in a chatbot interaction earns that trust. It also stops short of reporting the click-through comparison it sets up between chatbot, search, and social media users, so no finding here speaks to the "answer engine" referral-traffic worry the report raises. What the numbers support, at the strength a correlational survey allows, is that trust functions as a gate on deliberate AI use for news in a way it does not for incidentally-encountered social media — not that chatbot design is what produces the trust.