AI Search · 12 min read

Britons use AI but don’t trust it: what Ofcom’s 2026 data means for your marketing

Ofcom's 2026 data shows 54% of UK adults use AI, yet 56% trust AI-written news less. What the AI trust gap means for content, reviews and service.

A tea taster leans over a bench of white porcelain cups, spoon in hand, checking each blend in warm window light
On this page — 13 sections
  1. TL;DR
  2. Table of Contents
  3. What did Ofcom find about AI use and trust in 2026?
  4. Why do people use AI they don’t fully trust?
  5. Where do AI-first journeys actually end?
  6. What does the AI trust gap mean for your content?
  7. Why do reviews matter more in AI-led journeys?
  8. What does it mean for customer service?
  9. Our perspective
  10. Would your brand pass the check after an AI recommendation?
  11. FAQ
  12. Sources
  13. Recommended

Ofcom’s 2026 surveys point to a clear AI trust gap in the UK. More than half of adults (54%) now use AI tools and 21% used one for news in the past month, yet 56% say they trust an AI-written news story less than one written by a person. AI is where a lot of journeys now start. It isn’t where people decide who to believe.

You’ll recognise the pattern if you’ve watched how people actually buy. Someone asks ChatGPT for a shortlist, opens a few tabs, reads the reviews and, quite often, picks up the phone. The AI answer gets you into the room. Your brand has to do the rest. Below, we go through Ofcom’s three 2026 reports, add a B2B parallel from Gartner, and set out what it all means for your content, reviews, customer service and website.

If AI assistants are sending people to check you out, make sure there’s something credible waiting when they arrive. Our AI SEO team builds AI visibility and brand credibility as one piece of work.

Explore our AI SEO services

TL;DR

  • 54% of UK adults use AI tools (Ofcom, April 2026), rising to 79% of 16-24s.
  • 21% used AI for news in the past month, but 56% trust AI-written news less than human-written news (Ofcom, September 2026).
  • Ofcom’s telecoms review found people accept AI as a first line of support but want a human for complex problems.
  • In B2B, 69% of buyers prefer to validate AI insights with a sales rep (Gartner, May 2026).
  • AI-first journeys end on your site, your reviews or a phone call. Build AI visibility and brand credibility together.

Table of Contents

What did Ofcom find about AI use and trust in 2026?

Ofcom found that AI use in the UK is now mainstream, but trust hasn’t kept pace. Its reports published between April and September 2026 point the same way: people use AI to get started, then fall back on sources they already rate.

Adults’ Media Use and Attitudes 2026

Published on 2 April 2026, the Adults’ Media Use and Attitudes report found that 54% of UK adults now use AI tools such as ChatGPT, Copilot or Gemini. Younger adults drive it: 79% of 16-24s and 74% of 25-34s. Around one in eight AI users (12%) use these tools conversationally, rising to 19% of 25-34s. The survey covered 7,533 adults aged 16+, with fieldwork between September and November 2025.

News Report 2026

Ofcom’s News Report 2026, published on 15 September, asked about AI as a route to news for the first time. According to Ofcom’s summary of the findings:

  • 21% of adults used an AI app or service for news in the past month, with ChatGPT the most used at 13%.
  • Use is concentrated among younger adults and more affluent households.
  • 56% are less trustful of an AI-written news story than one written by a person.
  • UK adults come across around 12 different news sources a month on average, rising to 18 among 25-34s.

That last figure matters: people stack sources and cross-check as they go.

How trust compares across platforms

Ofcom’s Media Nations 2026 report, drawing on the same News Consumption Survey, shows where AI sits on trust. Only 51% of people who get news through search engines or AI rate them highly for trust, against 65% for broadcast TV and 66% for printed newspapers.

Share of users who rate each news platform highly for trust (UK, 2026)

Scale: 0 to 100%

Printed newspapers: 66%

Broadcast TV: 65%

Search engines or AI: 51%

Social media or video-sharing platforms: 39%

Source: Ofcom, Media Nations 2026 UK report (July 2026), citing the Ofcom News Consumption Survey 2026. Base: users of each platform for news.

The telecoms AI review, September 2026

In September 2026, Ofcom published a review of how AI is changing the telecoms customer experience. Its Technology Tracker found 53% of online adults had used a GenAI tool or an AI customer service chatbot in the past year, and 8% had used AI in relation to their phone, broadband or pay TV. Speaking to a person was still valued, Ofcom said, “particularly for complex or sensitive issues”. Its qualitative research (35 depth interviews in spring 2026) found AI was accepted for basic questions but challenged most on negotiating, decisions and complaints.

A buyer at a dawn fish market crouches by a box of fish on ice, pressing one to check freshness while holding a catalogue
The catalogue gets the buyer to the right box; the hands-on check decides the deal, much as people now check what AI tells them. AI-generated image.

Why do people use AI they don’t fully trust?

People use AI because it’s quick and good at the first pass. They hold back trust because they’ve learned it gets things wrong. So the journey doesn’t stop at the answer.

Think of it like satnav. You’ll happily follow it through an unfamiliar town, right up until it tells you to turn down a farm track with grass growing in the middle. Then you look out of the window.

The scepticism isn’t irrational. The BBC and EBU’s October 2025 study of more than 3,000 AI answers found 45% had at least one significant issue, and 31% had serious sourcing problems: missing, misleading or wrong attributions. If an assistant can garble a BBC story, it can certainly get your prices or service area wrong.

AI also turns up uninvited. Ofcom’s Online Nation report in December 2025 estimated that about 30% of searches now show AI Overviews. Even people who’ve never opened ChatGPT are getting an AI first draft.

The B2B version of the same story

B2B buyers behave the same way, only with bigger invoices. In a Gartner survey of 645 B2B buyers (fieldwork August to September 2025, published May 2026), 45% said they used GenAI during a recent purchase, mainly to research vendors and products. And 69% said they prefer to validate AI-generated insights with a sales rep.

The detail we like most: 51% of those buyers think they’re more likely to get misleading information from GenAI, and 49% say the same about sales reps. They check one against the other. If you sell B2B, the evidence on AI and shortlists is worth a read alongside this.

Where do AI-first journeys actually end?

For most considered purchases, AI-first journeys end on the brand’s own website, its reviews or a conversation with a person. The assistant shortlists; the brand gets checked; then someone decides.

Here’s an illustrative example. A homeowner in Chester asks ChatGPT for a reliable kitchen fitter nearby. It names three firms. She opens all three websites, scans the Google reviews and checks whether each firm actually covers her postcode. One site lists a service area that doesn’t include Chester. Another has no reviews newer than 2023. The third has a named fitter, recent photos of local jobs and a phone number at the top. Guess who gets the call.

The AI did the shortlisting; the brand evidence did the deciding. Two firms lost a job they’d already “won” in the AI answer. That’s why we tell clients to treat their website as the place AI-referred visitors come to verify. Every claim an AI makes about you gets tested there.

What does the AI trust gap mean for your content?

Your content needs to be checkable. If people are verifying what AI says, give them facts they can confirm and people they can identify.

  • Verifiable facts. Pricing, service areas, lead times and accreditations, stated plainly and kept current. Vague brochure copy can’t confirm anything.
  • Named authors. Google’s guidance on helpful content says it “strongly encourage(s) adding accurate authorship information, such as bylines”, and that of the E-E-A-T factors, “trust is most important”. A real person with a bio beats “Admin”.
  • Linked sources. If you quote a statistic, link the original. Readers check, and so do AI systems deciding what to cite. Our guide to earning LLM citations goes into how.
  • Dates. Undated stats look suspicious, because they often are. Date your articles and your data.
  • Consistent facts everywhere. Your name, address and services should match across your site, Google Business Profile, Companies House and directories. Assistants piece together who you are from all of it, the core idea behind entity SEO.
A stonemason's hands chisel a small geometric mark into the edge of a finished sandstone block in a dusty workshop
Masons signed their work so it could be traced, which is what named authors and linked sources do for your content. AI-generated image.

Why do reviews matter more in AI-led journeys?

Reviews are the independent check people run after an AI answer, so they carry more weight now. They’re also tightly regulated.

Since 6 April 2025, the Digital Markets, Competition and Consumers Act has banned fake reviews and concealed incentivised reviews outright. Businesses that publish reviews must take steps to make sure they’re genuine, and the CMA can fine firms up to 10% of worldwide turnover.

The CMA isn’t treating this as theory. After reviewing more than 100 business websites in July 2025, it found 54 could be failing to comply, often for having no clear policy on fake or incentivised reviews. In March 2026 it opened investigations into five businesses, including Autotrader and Just Eat.

  • Ask every genuine customer for a review, not just the happy ones you remember.
  • Reply to the bad reviews. A calm, specific reply often reassures more than ten five-star ratings.
  • Publish a short policy on fake and incentivised reviews.
  • Never buy, swap or gate reviews. A run of suspicious five-stars is exactly what a sceptical buyer spots.

What does it mean for customer service?

Use AI for the first line, but make the human easy to reach. That’s the clearest message from Ofcom’s telecoms work, and it applies well beyond broadband.

In its review, Ofcom said it should be “very clear” to customers when they’re dealing with a GenAI tool, and that it would be concerned if AI applications acted as “gatekeepers” blocking access to support. One of the people Ofcom spoke to summed up the frustration: with AI tools “you can go through all these questions and you end up back where you started.”

  • Label your chatbot as AI. Pretending it’s “Sophie from the team” damages trust the moment someone notices.
  • Put a phone number and email where people can find them, not three menus deep. AI assistants read your contact page too.
  • Offer a clear route to a person for complaints, cancellations and anything involving money.
  • Keep AI on triage and routine questions, where Ofcom’s participants were happy with it.

Our perspective

We think a lot of businesses are chasing AI visibility in the wrong order. They publish more pages and more FAQs and hope a model names them. Some of it works. But Ofcom’s 2026 data says being named is only half the job. If 56% of people trust AI-written news less than a person’s work, they’re unlikely to hand over a deposit just because a chatbot mentioned you.

So we build the two things together. AI visibility gets you onto the shortlist: clear entity signals, citable facts, coverage in the places assistants read. Brand credibility closes the deal: named experts, genuine reviews, a website that confirms everything the AI said, and a human who answers the phone. Skip the first and nobody finds you. Skip the second and you pay to be shortlisted, then lose at the check.

The good news is that the credibility work is often the cheaper half, because most of it is honesty and housekeeping. Our view, which we set out in why brand is the only moat left in AI search, is that the businesses people already trust will get the most from AI, not the least. For the full method, start with our Complete Guide to AI SEO and GEO.

Would your brand pass the check after an AI recommendation?

Not sure what people find when they click through from an AI answer? We’ll review your site, reviews and listings with you and show you where the gaps are.

Require AI SEO, GEO, or digital marketing and AI automation within your business? Book a free appointment with a senior consultant.

Talk to Castle

FAQ

Do UK consumers trust AI?

Mostly, they use it without fully trusting it. In Ofcom’s 2026 surveys, 54% of UK adults use AI tools, but 56% trust an AI-written news story less than one written by a person. Only 51% of people who get news through search engines or AI rate them highly for trust, compared with 65% for broadcast TV.

How many people in the UK use AI for news?

Ofcom’s News Report 2026, published in September 2026, found 21% of UK adults had used an AI app or service for news in the past month. ChatGPT was the most popular, used by 13% of adults for news. Use is concentrated among younger adults and more affluent households.

Does the AI trust gap affect B2B buyers too?

Yes. A Gartner survey of 645 B2B buyers, published in May 2026, found 45% used GenAI during a recent purchase, yet 69% prefer to validate AI-generated insights with a sales rep. Buyers use AI for research, then turn to people and brands they trust before committing.

How can a business build trust with people who found it through AI?

Make your website confirm what AI says about you. Publish verifiable facts, name your authors and experts, link your sources, keep business details consistent everywhere, collect genuine reviews and make a human easy to reach. Those are the things people check after an AI recommendation.

Sources