[SPONSORED CONTENT]
People are already using search engines, digital platforms and AI tools to think about money. This isn’t a future possibility. It is happening now. 40% of Brits are turning to ChatGPT, Gemini and Co-Pilot for financial guidance.1
The problem is not just that information can be wrong. It is that it can sound convincing. As Mercer highlights in its paper ‘AI consumer expectations and pensions’ confident seeming answers can be persuasive, but that doesn’t mean they are always accurate.
An analysis of Google’s AI overviews found they were accurate around nine times out of 10. At the scale Google operates, even a small error rate means millions of inaccurate answers. Equally concerning, more than half of the accurate responses were “ungrounded”, meaning the sources they linked to did not fully support the information given.2
A familiar problem, just on a bigger scale
People have always taken financial guidance from friends, family or the proverbial man down the pub.
What’s changed is the speed and confidence with which poor guidance can now be delivered. A tool can sound authoritative and make an uncertain answer feel trustworthy. Once that translates into action, the potential for harm is real. And because pensions are long term, the impact may take years to show itself and can build over time.3
Fraud is getting smarter too
This is not only about misinformation. AI is also making pension fraud harder to detect and easier to scale. Analysis from Pension Justice suggests £17.7 million was stolen in 2024 alone.4
This isn’t a hypothetical risk. It is live and present.
The point of action is where it matters most
Even if the conversation starts somewhere else, providers still have a role to play when a member takes an action, such as accessing pension savings. These are the moments that need the strongest checks, the clearest prompts and the most careful oversight.
As Matt Gosden, Head of AI at Mercer Master Trust, explains: “People are already getting guidance from outside sources. The problem is that consumers can’t always tell when the guidance is right and when it’s wrong. What matters for providers is what happens next. The goal is to make sure that, if guidance from any source is sub-optimal, the processes in place reduce the chance of harm when decisions are made.
“That is not a new responsibility. Providers already have processes in place to protect members when important decisions are being made. The difference now is that members may arrive with a firmer view of what they want to do, shaped by something they have read or been told elsewhere.”
Use technology, but keep control
The answer is not to try to control every outside source of information. That would not be realistic. Digital tools are useful when they improve access, speed and convenience. They can help members find information sooner, ask questions more easily and complete straightforward tasks with less friction.
But convenience is not the same as safety.
“Once a member moves from browsing to acting, the stakes change,” says Gosden. “At that point, expert judgement matters. Providers cannot assume the member has the full context, or that an AI tool has explained the risks properly. In many cases, they should assume this is not the case. Education is key, empowering customers to critically evaluate and recognise credible sources is essential.”
Make the decision point count
The question is not where the member’s idea came from. It is, does it make sense when it turns into action?
This is where providers still have a responsibility and where they need to focus. They cannot control every assertion a member sees online. But they can help shape the point at which decisions are made, making sure the risks are clearly explained, the right prompts are in place, and the opportunity to reconsider is built into the journey. That is the point that matters most.
To read more articles from Mercer, visit the content hub on Corporate Adviser – here.
____________________________________________________________________
1 Does AI give good financial advice? We asked it what to do with £16,000 | Money News | Sky News
2 How Accurate Are Google’s A.I. Overviews? – The New York Times
3 AI plan
4 AI Pension Fraud UK: £48,000 Lost Daily (2026)
