TL;DR:

  • Lloyds Banking Group is the first UK lender to introduce an AI tool that helps customers make investment decisions, piloting through Scottish Widows with a small customer group before a wider roll-out later this year.
  • Scottish Widows chief executive Chira Barua described the product as acting “like a satnav for investments” — offering guidance rather than regulated financial advice, a distinction with significant regulatory implications.
  • Lloyds is one of eight firms in the FCA’s newly-announced second AI Live Testing cohort; HSBC has told Reuters it is also studying similar AI deployments for customers with £15,000 to £37,500 (£20,000 to £50,000) in liquid assets.

The UK’s “advice gap” — estimated to cover millions of adults who cannot afford personalised financial advice and whose savings sit in cash rather than being invested — has been the FCA’s policy concern for close to a decade. Scottish Widows’ pilot is the first concrete attempt by a major bank to close it using generative AI, and it lands in the FCA’s structured observational programme rather than outside it, which is the model the regulator has been pushing publicly for at least two years.

Why the “guidance vs advice” distinction matters

Under UK rules, guidance is broad and generic; regulated advice must be tailored to an individual and triggers suitability, Financial Ombudsman and compensation-scheme obligations. The newly-created “targeted support” category sits between the two, deliberately lighter-touch than full advice, and it is the FCA’s principal tool for closing the advice gap. Scottish Widows’ AI-enabled targeted support product is among the first commercial tests of that middle layer, and its design — not the AI model quality — will determine whether targeted support becomes a viable mass-market category or remains a niche regulated product.

Risk signals already being flagged

Industry experts cited by Reuters warn that AI-driven guidance can amplify mistakes, mis-sell products and leave firms unable to explain outputs to customers or supervisors. The Bank of England is reportedly watching rollout patterns closely. A real-world test is whether Scottish Widows’ internal explainability and complaint-handling infrastructure can generate audit-ready rationale for each AI-produced recommendation — if it cannot, the regulator will have evidence to push harder on explainability requirements in the forthcoming Good and Poor Practice report.

Looking forward

HSBC’s parallel interest — specifically targeting the £15,000–£37,500 liquid-asset segment around ISA top-ups — suggests the UK retail banks are coordinating on AI wealth tools in a way they did not on robo-advice a decade ago. By end-2026, when the FCA publishes cohort evaluations, expect at least three of the four biggest UK banks to have live AI-guidance products. The differentiator will not be model quality — all will use similar frontier models — but rather the customer-data integration that lets the tool be useful without being over-specific. That is where UK incumbents hold structural advantage over neobank challengers.

Share this article