People who actually use AI chatbots for news trust them far more than people who don't — why the gap?

This explores why chatbot news users trust AI answers far more than non-users do. The gap is large. Among people who get news from chatbots, 44% trust that news, compared with 17% of non-users Does trust in AI chatbots drive news-seeking behavior?. Across the whole population, only about 20% trust chatbot answers, against 37% for news in general Why do AI chatbots gain news users but lose their trust?. The simplest explanation the corpus offers is self-selection: trust comes first and decides who adopts. Trust in chatbots predicts chatbot news use more strongly than trust in social media predicts social media use. That suggests people deliberately choose a chatbot as a news source when they already find it credible, rather than drifting into it.

Self-selection is only half the story. Other notes in the collection suggest that using a chatbot can also build trust, and not for reasons tied to accuracy. A focus group study found that ChatGPT earns trust through conversationality: quick, responsive, well-formatted replies trigger the same social instincts we use with people. Users rely on those cues instead of checking whether the answers are reliable Does conversational style actually make AI more trustworthy?. A related line of work finds that chatbots write in a register that signals expertise. Trust attaches to how an answer sounds, and users gradually stop searching and checking for themselves and simply rely on the system Does chatbot language style actually shape how much we trust it?. Non-users never get this exposure, so part of the gap may be trust that grows with use.

Time matters too. Long-term studies of personalized chatbots show trust rising with repeated use, but expectations rise with it. Each good interaction raises the baseline, so a later failure feels like a bigger betrayal Does chatbot personalization build trust or expose privacy risks?. A one-time survey snapshot can't show whether today's trusting users are on a steady climb or heading for a sharp drop.

This matters because the qualities that feel trustworthy can work against reliability. Research on training AI to be warmer and more empathetic found that it increased errors in truthfulness and made the models worse at resisting disinformation, by up to 30 percentage points. The effect was strongest when users expressed false beliefs Does empathy training make AI systems less reliable?. For news, that is an uncomfortable combination: the friendly style that wins users over may make a chatbot more likely to agree with a reader's mistaken view of events. There is a counterweight. 42% of chatbot news users click through to the original sources Why do AI chatbots gain news users but lose their trust?, so a large share of these trusting users still check.

The corpus doesn't settle how much of the 44-vs-17 gap is selection and how much is trust built through use. No note follows the same people from before they adopted chatbots to after. The useful takeaway is that, in this research, chatbot trust mostly tracks how a chatbot talks, not how often it's right. That makes the gap a measure of how persuasive the interface is, not proof that users found the news more accurate.