“I’ll Just Read The AI Summary” Became My Whole Morning
Lately the news arrives pre-chewed. You paste the report into a chat window, ask for three bullets, and only open the original if something in the summary snags.
I do this constantly. Long PDFs especially — “give me this in three lines” has quietly become my default reading posture.
But a small voice shows up right after. How much of this do I actually believe? Checking whether the source was right, or whether the model padded it a little, costs more effort than the summary saved.
And that “handy, but I can’t quite trust it” feeling turns out not to be mine alone. It shows up in consumer data across dozens of countries, and it points somewhere very specific: at your brand name.
So let’s line up three surveys — Reuters, Edelman, Gartner — and see where trust actually lands.
How Many People Really Moved To AI For News
The starting point is the Reuters Institute at Oxford, whose “Digital News Report 2025” came out in June 2025. It tracks how people around the world get their information, every year, at fairly serious scale.
And the generative-AI numbers moved fast:
- People who’ve used generative AI at all: 40% to 61%
- People using it weekly or more: 18% to 34%
So in one year the world went from “I’ve poked at it” to “it’s part of my week”. On news specifically, summarising articles is the most wanted use at 27%, ahead of translation (24%) and better recommendations (21%).
Which sets up the interesting part.
Handy, Yes. Trusted, Not So Much.
The same Reuters survey asked what people make of news produced by AI, and the answers pull in two directions at once.
AI-made news reads as cheaper to produce (+29) and faster to arrive (+16). Fine, nobody’s arguing. But on the things that decide whether you act on it, the scores go negative: less transparent (-8), less accurate (-8), and less trustworthy at -18.
So people take AI as a useful tool and squint at it as a source. It’s a bit like the fridge — obviously useful, which doesn’t mean you trust the container that’s been sitting at the back of it since spring.
Here’s the part marketers shouldn’t skip, though. Reuters points out that trusted news brands are working as the single most important way people fact-check what they see. The more AI-shaped information there is, the more often somebody stops and asks “hang on, who’s actually saying this?” — and a name they already trust is what answers.
Trust Splits On Whether You’ve Ever Used It
Edelman’s “Trust Barometer”, published in January 2026, comes at that split from a different side: 28 countries, around 34,000 people.
Trust in AI in the US sits at 32%. Generative AI platforms even made the list of the top five things that damaged trust over the past five years — 37% of respondents picked them.
But look at who’s answering. Between people who’ve used AI and people who haven’t, trust differs by 26 to 46 points. The ones who’ve touched it trust it more, which means the never-tried-it crowd is still thick on the ground.
So “distrust of AI” is too blunt a phrase to plan around. There are two audiences in there, and one message won’t reach both.
Half Of Consumers Prefer Brands That Skip The AI
Now the survey that lands closest to the actual job. In March 2026 Gartner published a study of 1,539 US consumers, and the headline number has some punch: 50% said they prefer brands that don’t use generative AI in customer-facing messages, ads and content.
The verification behaviour underneath it is just as telling:
- 61% frequently doubt whether the information they use for everyday decisions is reliable
- 68% frequently doubt whether the content in front of them is genuine
And as of late 2025, only 27% judge whether something is true on gut feel. People have stopped trusting their instincts and started checking.
None of which is a blanket rejection of AI, mind you. A separate BCG survey found more than 60% say they trust generative AI highly. Both things sit there at once: the tool is fine, and the brand’s own voice arriving with AI mixed into it is what makes people flinch.
Conclusion: Treat AI Use As A Trust Question, Not A Disclosure One
Line the three surveys up and the picture is pretty clear. Far more people now meet information through AI. Trust in what AI produces stayed low, so people go hunting for the source. And the more AI spreads, the more often the answer to “who’s saying this?” has to be a name they recognise — which is a backhanded but real argument for making sure the models can name you at all.
So the takeaway for a marketer or brand lead isn’t a rule about disclosure. It’s that AI use is a trust design problem.
Quietly hiding the AI is a bad move, and bannering “Now with AI!” across everything is a bad move too. Half of consumers are wary of AI-flavoured communication, and the “used it and warmed to it” group and the “keeping my distance” group don’t hear the same sentence the same way. What matters isn’t use-versus-hide — it’s whether a reader can still tell who stands behind the claim.
Two things pay off, then. Keep whatever you make with AI checkable: name the primary source, name whoever reviewed it. And on some regular cadence, look at what ChatGPT and the rest actually say about your brand when someone asks, because that’s the context your name turns up in, and the surveys say people are now going looking for exactly that. If you want to know where your own customers sit, the Reuters, Edelman and Gartner reports are all worth reading first-hand.
I think as AI becomes the front door to information, the thing that decides the outcome is the unglamorous pile of “you can believe these people”. That’s the part you can’t hand to a model.
Sources
- Reuters Institute for the Study of Journalism (2025), “Digital News Report 2025”, reutersinstitute.politics.ox.ac.uk
- Edelman Trust Institute (2026), “2026 Edelman Trust Barometer”, edelman.com
- Gartner (2026), “Gartner Marketing Survey Finds 50% of Consumers Prefer Brands That Avoid Using GenAI in Consumer-Facing Content”, gartner.com
