Why do AI chatbots gain news users but lose their trust?

The Reuters Institute's Digital News Report 2026, an annual survey run across 48 markets, finds that weekly use of AI chatbots such as ChatGPT, Perplexity, and Google Gemini for news "increased year-on-year from 7% to 10%," which the report calls "fast rather than explosive growth." Adoption is concentrated among the young and the already-engaged: 16% of under-35s report using chatbots for news, and while South Korea, Greece, and Spain saw usage double, the USA, the UK, France, and Germany "reported no increase... over the year." This sits inside a wider "platformisation" shift in which social media and video networks are, "for the first time," ahead of all other news sources at 54% of audiences globally, while trust in news overall has fallen to "the lowest level... recorded since we started to measure trust in 2015 (37%)."

The report treats adoption and trust as separate, divergent lines. "Trust in answers from AI chatbots is 20% globally (relative to trust in news overall at 37%)," and in the UK — the market with the lowest chatbot-for-news usage (4%) — trust in chatbot answers is also lowest at 6%. The report attributes part of the gap to "the tone of media coverage about AI generally" shaping perceived trustworthiness, independent of any individual's chatbot use. On the publisher side it frames the stakes as a "Google Zero" question: whether audiences satisfied by a chatbot's answer stop clicking through to the original reporting. Here 42% of chatbot news users "say they always or often click through... but there are wide variations by market," from 56% in South Korea to 26% in Denmark — and the report flags this as self-report, noting "this may be an area where there are differences between what people say they do and their actual behaviour."

Against Does chatbot language style actually shape how much we trust it?, this report supplies population-scale numbers for a handoff that note describes rhetorically: chatbots are gaining ground as a news intermediary, yet trust has not followed usage at anything like the same rate — the register of expertise that paper describes does not appear to be winning open trust once measured across 48 markets. It also parallels Do AI writing tools improve online discussion or degrade it?: both find a split rather than a single tradeoff — one thing growing (chatbot use, AI-assisted output) while a separate judgment (trust in the answer, perceived quality of the content) fails to rise with it, or falls.

As a self-report survey, this excerpt cannot say whether low chatbot-news trust reflects direct experience with specific chatbot answers or spillover from general anxiety about AI, nor whether the stated 42% click-through rate holds up against platform log data — the report itself concedes the behavior/attitude gap it otherwise documents for news consumption generally. What the figures support, at the strength given, is narrower: AI chatbots are becoming a modest and unevenly adopted news channel, used disproportionately by an already-engaged minority who say they trust its answers less than news overall and who, even by their own account, mostly do not return to the source once they have one.