Tech-savvy isn’t necessarily a trait people associate with regulators. Yet, slowly but surely, the Financial Conduct Authority has carved out a reputation for being pretty forward-thinking.
Natural language processing and triage
It’s long been ahead of the curve with the likes of its financial advice sandbox, allowing planners to try out new kit without fear of reprisal. Now, how it chooses to use artificial intelligence could be a watershed moment for advisers. A recent Freedom of Information Act request shows how far the watchdog has come.
“The FCA is committed to using data and technology to make it a more effective and efficient regulator,” it says. “To support this we have invested in data science capability and capacity across the organisation to understand how the use of data science and advanced analytics can be used to bring new insight.”
First up, the regulator says some of its staff are using natural language processing to prioritise which documents to review and analyse so-called ‘unstructured text’ faster and more effectively.
Supervisors currently spend a lot of their time sifting through documentation and management information from firms like board minutes, committee papers, compliance reports and the like. The FCA has already shown it can successfully use its reporting platform to throw up outliers in areas like pensions advice. Essentially if it sees a spike, the system automatically generates a red flag and you may well get a visit from our dear friends at Stratford, or at least a nice email asking you to explain the pattern it’s picked up. Rolling this out more widely can only help weed out the rogues dragging the sector down.
Improving the speed of triage could also be a leap forward. Done right, this use of AI could see advisers’ regulatory bills reduced – after all every firm that fails is a cost on the Financial Services Compensation Scheme, which is then passed on to the wider profession. If AI was able to pick out suspicious actions faster, then you wouldn’t have as many clients go broke, even if you couldn’t stop the original mis-selling, so you don’t get as big a hit on the FSCS that everyone else pays for.
Deep learning and friendlier regulation
So much for what’s going on right now in regulatory technology. But the future that should excite financial planners as well. Because the FCA says it is working on some proofs-of-concept that could also show the power AI has to create much friendlier regulation.
These include using deep learning to search documents not just for words but for context instead, detecting images within documents, and supervised machine learning to prioritise workloads.
A lot of this seems unremarkable at first glance. But just think of the implications. Imagine if the FCA could search firms’ records not just for literal phrases that might hint at villainy – “hey boss, we ripped a client off with this DB transfer, I shouldn’t tell anyone right? – but also for internal inconsistencies or for where the meaning of text might be a compliance concern when set in the proper context.
Recognising images sounds pretty ordinary as well, until you remember that whistle-blowers often send over screenshots or other pictographic evidence for the FCA to churn through. If the watchdog can do that so much easier (let’s say by picking out what relates to say mini-bonds, or the next scam) it could help the watchdog regain the good faith all those advisers who have reported malpractice to no affect in the past.
More advisers would come forward too because they trust their evidence might actually be used for something, and planners would not just sit on what they know for fear it would just get lost in the regulatory quagmire.
Let’s imagine these conceptual proofs then make their way to the FCA’s authorisations team. One of the big bugbears of advisers – that getting permission to set up a new firm or launching an innovative service takes an absolute age – might not disappear entirely. But AI could certainly cut into wait times if the machines help tell the regulator which firms really are worthy of making it through its net.
Promoting AI for advisers
That’s not the end of it, either. While it increases its use of AI internally, the FCA is also trying to help advisers themselves understand what it could mean for their businesses.
It was involved in creating guidance issued by the International Organization of Securities Commissions in late 2021 specifically targeted at intermediaries and asset managers using artificial intelligence and machine learning. The goal? To balance the potential gains of lower investment costs and more time spent with clients against the risk that consumers could be harmed in new ways.
That guidance was among 12 other speeches or papers on the topic of AI the FCA pointed to in the FOI response. So whether advisers are aware of them or not, the influence of the regulatory robots is only set to grow.
Photo by Shironosov on Canva
