Research·3 min read·Reuters Institute for the Study of Journalism

AI Chatbots Now Deliver Weekly News to 1 in 10 Adults

The Reuters Institute’s 2026 Digital News Report finds weekly chatbot news use up to 10% globally — younger, highly engaged, and rarely clicking through to the source.

AI Chatbots Now Deliver Weekly News to 1 in 10 Adults
Share:

The way the world finds out what is happening is shifting toward AI assistants — but more slowly than the breathless predictions of a year ago suggested. The Reuters Institute for the Study of Journalism released its 2026 Digital News Report this week, surveying nearly 100,000 people across roughly 50 markets, and one number anchors the AI story: 10 percent of respondents said they used a chatbot such as ChatGPT, Google Gemini or Perplexity to get news in the past week, up from 7 percent a year earlier.

That three-point rise reads as steady rather than explosive, and it is strikingly uneven. The fastest growth came in Asia, Africa, Latin America and parts of Southern and Eastern Europe, while several of the largest Western markets — including the United States, the United Kingdom, France and Germany — recorded no meaningful increase at all. For all the attention chatbots command, just 1 percent of people named one as their main source of news.

The behavior skews young. About 17 percent of the youngest age group used a chatbot for news in the past week, more than three times the 5 percent recorded among the oldest. The sharpest year-over-year jump came from 25-to-34-year-olds, up four points. Usage also clusters among the most engaged: 18 percent of self-described "news lovers" reached for a chatbot, against 7 percent of lighter news consumers.

Trust tells a more complicated tale. Across the general population, only 20 percent said they trust AI to produce accurate news, well below the 37 percent who trust traditional news overall. Yet among people who actually use chatbots for news, trust jumps to 44 percent — a familiarity effect that worries researchers as much as it reassures the AI labs, because confidence is growing fastest among the people most exposed to the tools' mistakes.

The figure that should alarm publishers is click-through. Only 4 percent of respondents said they always or often follow a link from a chatbot to the original source, compared with 19 percent from search engines and 17 percent from social media. South Korea topped the chatbot list at 8 percent; the UK sat at just 1 percent. As AI summaries absorb more of the reading, the referral traffic that funds journalism is leaking away — and the answers themselves carry no byline and little accountability.

Asked why they turn to chatbots, people cited depth and explanation (42 percent), speed (39 percent), the ability to summarize complicated stories (36 percent) and translation (33 percent). The most common tasks were asking follow-up questions (42 percent), getting the latest news (35 percent) and summarizing (34 percent). The report frames an "unsettled time" for the news business: AI is not yet the front door to news for most people, but it is quietly becoming the lens through which a growing, younger, highly engaged minority reads the world — and that minority rarely clicks back to where the reporting began.

Comments

Share your thoughts. Be kind.

0/2000

Loading comments…

Related Articles

RESEARCH · OPENAI JUN 17 An AI chemist cracked a stubborn drug reaction. GPT-5.4 and Molecule.one’s Maria Lab pushed a low-yielding Chan-Lam coupling much higher. 88% boronic acids improved 83% sulfonamides improved 8 / 14 validated reactions more than doubled GPT-5.4 proposed and ranked experiments; Molecule.one’s Maria Lab ran them. Human chemists steered the work and validated the result. Start to finish: about 2.5 months. BITSMINDS.COM Source: OpenAI · Molecule.one
Research

OpenAI’s “AI Chemist” Improved a Reaction Drug Makers Had Nearly Given Up On

OpenAI's LifeSciBench Puts AI Through a 750-Task Life-Science Exam — Top Model Passes Just 36%
Research

OpenAI's LifeSciBench Puts AI Through a 750-Task Life-Science Exam — Top Model Passes Just 36%

AI SAFETY · OPENAI JUN 16 OpenAI now rehearses a model before it ships. Deployment Simulation replays 1.3M past chats through a new model to forecast misbehavior. STEP 1 Replay recent production chats STEP 2 Regenerate the reply, new model STEP 3 Grade it for misbehavior STEP 4 Estimate the deployment rate Median error 1.5x · caught “calculator hacking” in GPT-5.1 before release Built from ~1.3M de-identified conversations, Aug 2025 to Mar 2026. BITSMINDS.COM Source: OpenAI · MarkTechPost
Research

OpenAI’s “Deployment Simulation” Replays 1.3 Million Real Chats to Catch a Model Misbehaving Before Launch