Press

8 January 2026

For immediate release

70% of users accept AI-generated content without checking sources

Research from communications consultancy Quietroom reveals how AI-powered search tools struggle with pension scheme content, leading to incorrect information being provided to members.

In the UK, 92% of people search the internet using Google. Google now generates AI overviews for approximately half of all searches. Research studies show that as many as 7 in 10 users take the information in these overviews at face value. They don't check where the information has come from, and they don't view the webpages listed in the sources.

Simon Grover, Director at Quietroom, said

“Members are no longer reading what their scheme has written – they’re reading what AI tools serve up, which may or may not be accurate. And they’re asking AI to give them key points and what decision they should make."

Quietroom tested OpenAI's new Operator agent on UK pension websites. In one test, Operator was asked to figure out when a deferred member can retire from a scheme. The website used accordions (boxes that expand when clicked). Operator tried to interact with them but failed to properly read the information contained within them. This meant it gave the wrong answer about when a member can retire.

In other tests, Quietroom asked ChatGPT and Google simple questions about a pension scheme – questions that they had already checked were answered somewhere on the scheme website. But because of the way the content was organised, the AI tools couldn't find it. And instead of admitting they couldn't find it, they made up answers based on other scheme websites. As the schemes operated differently, the answers were wrong. These results happened repeatedly across various pension scheme sites.

Quietroom's audits have found AI serving answers from a different scheme with the same initials. They've seen AI assistants offer to perform complex calculations for members – and then get them wrong. They've also seen AI direct members not to the scheme or its administrators but to third parties – including disreputable financial advisers.

The testing found that if content is not where users expect to find it, both humans and AI agents struggle. Large language models (LLMs) that power AI tools have limitations that echo human readers. Humans can typically hold 5 pieces of information in their working memory. For some this might drop to just 3. LLMs have a similar constraint called the 'context window'. Research shows that, as a result, LLMs can miss information that is hidden in the middle of documents.

AI assistants cannot distinguish between different cohorts of members – someone who joined in 1990 might have different benefits to someone who joined in 2005, but the AI treats them the same. When scheme content uses formal or technical language, AI tries to simplify it – and risks distorting the meaning. Schemes sitting on decades of communication history reflecting benefit structures, accrual rates and pension increase policies that have changed multiple times face particular challenges, with outdated content being treated as current.

"The solution isn't to write for robots, but to write better for humans," notes Grover. "Our research shows that AI does a much better job accurately summarising or explaining content if that content is already clear, consistent, well-structured and in short sentences."

When a user gets an answer from a search engine without following up with a visit to a website, it's called a 'zero click' search. Zero click searches are on the rise. This means content may no longer be seen the way it was intended. In fact it might not be seen at all.

"The more 'gappy', complex, and difficult it is to use your content, the more likely it is that generative AI will give your members incorrect answers," states Grover.

The FCA has made clear that Consumer Duty applies to AI-driven customer interactions – and that firms remain accountable for the outcomes. Members making decisions based on wrong information leaves schemes accountable for the content that led them astray.

Notes to editors

Quietroom is an insight-led communications consultancy specialising in making pensions, investment and insurance accessible.

The research involved testing AI search tools including Google AI Overview, ChatGPT search and OpenAI's Operator agent on UK pension websites.

Contact information

Quietroom
Email: simon.grover@quietroom.co.uk
Tel: 07951 290732
Web: quietroom.co.uk

Ends

10 June 2025

Quietroom, the insight-led communications consultancy, has carried out research that suggests pension scheme members are more open to UK investment than the pensions industry itself. While the industry has concerns about government pressure to invest more in the UK, members are surprisingly positive about the idea.

The research, conducted through 8 random interviews in a London park, found members relaxed about UK investment, while the industry worries about fiduciary duty and political interference. The findings are captured in a video available at on the Quietroom website.

Simon Grover, Director at Quietroom, said

"The pensions industry has been having technical conversations amongst ourselves about UK investment. We forgot to do something simple: we forgot to ask members what they think. What we found challenges the industry debate. Once again, this research shows the value and power of testing your ideas with real people."

Nearly everyone interviewed was positive about UK pension schemes investing more in the UK. A barrister trainee thought it "makes sense" for government to use pension pots domestically. A university employee said it would "obviously be good if we were investing in the UK". An academic was clear: "I think it's a good idea. Definitely."

Members care about returns but showed flexibility unlike the industry's technical debates. They used phrases like "as long as it doesn't make too much difference" and "so long as I don't have a huge drop". They weren't demanding maximum returns at all costs but were open to balance.

Participants immediately grasped benefits that rarely feature in industry discussions. They talked about supporting British jobs and growing the UK economy. One respondent suggested UK investment could create a virtuous cycle where more investment improves returns.

"Just an hour in a park taught us something crucial," said Grover. "There's a substantial gap between what the industry worries about and what members think. If we don't understand what they think, how can we serve them properly?"

The findings suggest the industry may be so deep in technical debates that it's lost sight of what members want. Members want decent returns but also want their money doing something worthwhile. They're willing to accept trade-offs the industry assumes they'd reject.

Quietroom works with leading organisations in financial services, pensions, and insurance to make complex information clear, helping people make better-informed decisions about their money and their future.