Is regulation keeping up with AI chatbots?
Alison Gay of The Lang Cat considers the safeguards around artificial intelligence advice.
You’ve just had a big win on the Premium Bonds.
Once you’ve spent some of it on a Big Night Out you feel it would be a good idea to save or invest the rest. But you’re not sure where to start.
Or you’re approaching retirement, you’re vaguely aware of having pensions from various jobs but you don’t know what to do with them or whether you’ll have enough to live on.
Or you want to work out a savings plan for your children that will help them fund university studies.
Until now you had, essentially, two routes to follow for each of these situations. You could plough on by yourself, using whatever skills and information you already have and researching options. You might get some help from generic financial information, or from friends or relatives. Or you could go down the route of regulated advice and guidance, seeking help from firms governed by specific rules, in the knowledge that if anything went wrong there was a whole structure of complaints and redress procedures to protect your interests.
But now the world is, for better or worse, different. Millions of people are already using general purpose AI tools like ChatGPT, Claude or Gemini for what is, in effect, financial advice. You can put in your whole financial history and get a plan of action generated for you, apparently neatly tailored to your needs and written in a friendly style.
Unfortunately, you have no way of telling if it’s factually accurate or full of misinformation and AI hallucinations. You might get a response that says ‘I’m not a financial adviser, so treat this as information to help you decide, not a recommendation’. Or you might not. And if you act on it and it all goes horribly wrong, you have no recourse to redress at all.
The question here isn’t about whether the ‘advice’ received from an AI chatbot is comparable to the service you get from a regulated adviser (although at the lang cat this is something we’ll be testing at our next event in Leith in October). It’s about what the regulators, and the government’s, thoughts are about putting safeguards in place.
Is regulation keeping up?
In July a Financial Services AI Adoption Plan was published by the government’s ‘AI Champions’ Harriet Rees of Starling Bank and Dr Rohit Dhawan of Lloyds Banking Group.
The report came up with ten recommendations to speed up safe AI adoption covering regulatory clarity, the boundary between regulated advice and AI-generated guidance, operational resilience, skills, and agentic payments.
Around the same time the Financial Conduct Authority published its Mills Review which looked at how AI could reshape retail financial services by 2030 and beyond and also made recommendations to the FCA on how it could respond.
Both reviews agreed on some things. The AI cat is already out of the bag – people are already using it to make decisions. Neither wants a whole new rulebook put in place to deal with this, but it needs to be made clear to people what risks they are running by using the new technology.
The Adoption Plan sees the solution as more industry-led through more disclosure, making it clearer what is and is not regulated, and trusting the market to come up with solutions. The biggest point of difference is that the Mills Review does not trust the market to deliver on its own but proposes the development of a trusted public AI advice service.
Views of the advice profession
What about the advice profession, which as a sector relying on human relationships and interaction might be expected to push back on the use of AI?
At the Lang Cat we’ve been asking advisers their opinions on the subject for several years in our State of the Advice Nation reports. When we asked this year if AI was a barrier or threat to their role or to the profession, the dominant response was that it’s an opportunity rather than a threat. A smaller group view AI as both, recognising its benefits while acknowledging associated risks. Only a small minority see AI primarily as a threat.
Firms themselves are already using the technology, most commonly to help with administrative and data handling tasks. In the space of two annual research waves adoption has moved from a minority pursuit to a majority practice. When it comes to comfort levels with AI-generated recommendations for client-facing advice though, support drops sharply.
Firms understand what AI can do, but also understand that decisions need to be explained, evidenced, and justified when challenged. The profession understands that accountability does not disappear simply because a tool is involved; if anything, it becomes more complex.
In other words, when asked about regulation specifically, the advice profession wants the regulator to catch up, rather than crack down. But there’s a lot of work to do to make sure that people understand what risks they’re running when they fire up the robot to ask for help.
For now, accountability and decision-making remain firmly human responsibilities. Maybe stick to using it to plan that Big Night Out, rather than making life changing decisions about what to do with the rest of your winnings, at least for the time being.