People Use Chatbots for News. Barely Anyone Checks the Source.

Ten percent. That's how many people worldwide now use AI chatbots for news every week.

A year ago, it was seven percent. The year before that, four. Whatever ceiling exists on this behavior, we haven't found it yet.

The Reuters Institute 2026 Digital News Report buried one of the more interesting data puzzles of the year: ten percent of people globally turn to AI chatbots for news. But only twenty percent of people globally trust news from AI chatbots.

Do the math. The people getting news from AI are a subset of all people. So either chatbot users trust it more than average, or a bunch of people are getting their news from a source they don't believe. Turns out, it's a bit of both.

Among people who actually use AI chatbots for news, trust jumps to forty-four percent. Non-users trust it at seventeen percent. That gap is doing a lot of work. The more you use it, the more you trust it. Which is either a story about familiarity building confidence, or a story about repeated exposure eroding skepticism. Depending on your mood.

Here's the number that matters most: four percent.

That's the share of chatbot news consumers who click through to an original source. Search sends nineteen percent to source material. Social media sends seventeen percent. Chatbots? Four.

For comparison, four percent is also roughly the error rate on a coin flip if you're doing it wrong.

When Reuters dug into why that four percent does click through, the answer was revealing: they're mostly doing it to verify what the chatbot told them. Not to read more. Not because they found the story interesting. Because they're suspicious of what they just read. The chatbot is the fact-check prompt, not the fact.

This would be more reassuring if the other ninety-six percent were also checking.

Reuters framed this charitably. Maybe chatbot news is "complementary," a bridge to real sources. Only one percent of people call AI their primary news source. Most people are mixing it in. That's the optimistic read.

The less optimistic read involves a study published earlier this year finding that AI chatbots don't just spread misinformation: they can actively reinforce it. The conversational format validates and builds on whatever the user already believes. You come in with a half-formed idea. The chatbot engages with it thoughtfully. You leave more confident.

So: a tool that people use for news, that most people don't fully trust, that almost nobody fact-checks, and that tends to strengthen existing beliefs when pressed. What could go wrong.

The broader context doesn't help. Reuters also found that trust in news overall is at a record low. This is the room chatbots are walking into. People don't trust newspapers. They don't trust television. They definitely don't trust social media. And now they're turning to AI, which they trust even less, but which at least gives them a complete sentence instead of a doom-scrolling notification.

The youngest users are adopting fastest. Seventeen percent of the youngest age cohort uses chatbots for news weekly, versus five percent of the oldest. If you think trust in news has been declining for a generation, consider that the generation coming up is getting its information from systems with a twenty percent trust baseline. The numbers will only go up from here.

There's a version of this that works out fine. AI gets better at citing sources. People get better at verifying. The click-through rate goes from four percent to something respectable. The hallucination rate falls. The tools mature. If you want to understand where today's major chatbots actually stand, comparing them side by side helps calibrate what you're actually trusting.

We're not in that version yet.

What we're in is the version where ten percent of people are already using this, the number is climbing, the source-checking rate is somewhere between token and theoretical, and the most engaged news consumers are the ones doing it most. Reuters calls those people "news lovers." I'm sure they are. I'm also sure that loving news and getting good news are not the same thing.

The question for the industry is whether the four percent who check sources are an early-adopter vanguard or whether they're always going to be four percent. I have a guess. It is not optimistic.