Protection advisers have overwhelmingly backed the Financial Conduct Authority’s (FCA) call for consumer AI services such as ChatGPT, Gemini and Claude to include disclaimers or warnings to help consumers understand the limits of financial information and advice generated by their models.
The call coincided with the regulator’s release of its AI in financial services: good practice and areas for improvement report this morning which shared examples from the regulator’s industry engagement, supervisory work and its AI Lab initiatives to support the safe and responsible adoption of AI in financial services.
Health & Protection spoke to advisers about the regulator’s report and found they overwhelmingly backed the FCA’s call to add a warning.
Inaccurate information
Alan Lakey, director at Highclere Financial Services at CI Expert, told Health & Protection: “The current AI services are notorious for providing inaccurate information. I chuckle when I correct their misinformation and they apologise and thank me for the correction. Disclaimers should be on all of these.”
Didn’t have a clue
Karla Edwards, owner of The Protection Parent, agreed, adding she discovered the limitations of the technology just this morning.
“I went on ChatGPT this morning to see if I was a customer and what it would recommend and it didn’t have a clue,” Edward said.
“When asked about global treatment, it was telling me to go to a provider, but they can’t help as you have to get it through a regulated adviser.
“I kept probing it and it wasn’t giving the right information and I thought there would be people using this for information and you know that the information’s not right. It’s scary to think people are trusting over something so important.”
No-one to fact-check
For Naomi Greatorex, owner of Heath Protection Solutions, the key issue is that in these situations customers don’t get a second opinion.
“I think it’s important to know where the information’s coming from and its limits always.” Greatrorex maintained.
“It’s difficult because you’ve got no-one to fact-check it and you do have that issue of it being wrong.
“We get it occasionally, but not as much, because of where our clients come from, because our clients come from a professional introduction, but I can see from some areas of protection advice, it could really affect their line. If people are getting information from online brokers, I could see that could be a blurry line that needs to be checked.”
Same rules apply
For Rhys Schofield, protection adviser at Peak Mortgages and Protection, while the same marketing rules should apply, new regulations may be required.
“If I have to stick a risk warning on a social media post, it stands to reason the same marketing rules should apply to tools people are using for actual guidance – especially given AI can provide incorrect information,” Schofield said.
“That being said, I’ve started to get enquiries recently because ChatGPT is telling clients to contact me for advice so it’s clearly doing something right.
“Given the FCA guidelines on promotions and adverts were written in 2015, and last updated in 2024, perhaps it’s time for a full revamp of the rules so they are fit for the world we live in.
“Presently, too much is open to interpretation.”
Charlotte Rogers, protection adviser at Radcliffe & Co, also backed the FCA’s call for disclaimer.
“AI is not 100% accurate yet and clients could easily be given incorrect information without AI knowing the full story,” Rogers said. “AI is not going to know the client’s medical history or attitude to risk to name a couple of areas. The public could easily interpret the information as accurate and believe they need/don’t need a particular product or service when in fact the opposite may be true.”
Applying critical thinking
But Kristian Breeze, director at Ascend Broking Group, maintains that he is sceptical whenever the answer to a new technology is more warnings, more regulation and more intervention.
“AI is a tool, not an adviser, and consumers should be encouraged to apply the same critical thinking they would to anything they read online,” Breeze said.
“Some customers undoubtedly place too much trust in AI-generated answers, but that is not unique to AI. People have always taken information from websites, forums and social media as fact. The real issue is digital literacy, not a lack of disclaimers.
“AI can sometimes produce outdated or inaccurate information on topics such as underwriting terms, policy exclusions or insurer appetite, but the market is moving quickly to correct these shortcomings. Excessive oversight risks stifling innovation in a technology that has the potential to make financial information more accessible and understandable for consumers.”
Proportionate warning
Justin Harper, chief marketing officer at LifeSearch, an advice that firm that has embraced the technology, said the company supports clear, proportionate warnings that help consumers understand the limits of AI-generated information, what protections apply, and who they can turn to if something goes wrong.
Huge opportunity
But Harper added that AI presents a huge opportunity for our market.
“One of the biggest challenges in protection is that many people have never even considered their needs,” he continued.
“The bigger risk for our market isn’t that AI gets every answer right… it’s that millions of people never ask the question in the first place. If AI helps bring conversations about life insurance, critical illness cover, income protection and financial resilience into everyday searches and life moments, that’s a good thing. It has the potential to make protection more visible in people’s experiences, more relevant and easier to understand.
“AI is changing the nature of our conversations with customers. Powered by AI, they are rapidly becoming both the most informed and the most misinformed generation at the same time – those two things can coexist.”
Overconfidence in the technology
Harper added that he sees the challenge as not that AI lacks information, but that people can sometimes place too much confidence in information that sounds authoritative.
“Protection is rarely a simple numbers exercise,” he continued. “It involves health, family circumstances, future plans, affordability, underwriting considerations and the people who matter most. For us, we find those nuances matter.
“We don’t see this as AI versus humans, rather as AI plus human expertise. AI can be brilliant at helping people understand risks, explore scenarios and ask better questions. But there remains a real need for a specialist intelligence layer that is transparent, accountable, grounded in real-world expertise and, where appropriate, regulated.
“In time – and it may not be that far away – I suspect we’ll see those specialist intelligence layers sitting alongside and powering AI ecosystems, rather than relying solely on general purpose curated models.”