The Financial Conduct Authority has put a number on the supply side of Britain’s new investment-support market. Nine firms have been authorised to provide Targeted Support, its director of consumer investments said on 18 September. It has not put a comparable number on reach.
That distinction matters because the competing channel is already in people’s hands. An FCA survey published in August said four in five less-experienced investors had used artificial intelligence for help with investing. Yet the regulator has not published the denominator for that subgroup, a questionnaire or tables showing whether “help” meant decoding jargon, researching a company, asking for a recommendation or completing a trade.
Permissions and survey responses therefore cannot be placed on opposite sides of a market-share chart. What can be tested is whether the regulated channel has become visible, usable and measurable since its April launch.
The AI figure has a narrow selection boundary
The FCA conducted the survey via the Attest platform on 24 July. Its 666 respondents were UK residents aged 18 to 40 who either owned investments or would consider buying them within 12 months. It was not a survey of all UK adults or even all people in that age band.
Within that selected sample, 56% said they trusted AI tools. Some 44% thought AI-generated financial information was regulated, 38% considered it acceptable to make an investment decision solely from AI output, and 32% thought the Financial Services Compensation Scheme or Financial Ombudsman Service would compensate them if AI advice went wrong.
Those findings expose confusion about protection. They do not establish that four fifths received a product recommendation, acted on one or funded an account. The unpublished subgroup size also prevents readers from converting the headline ratio into a reliable count. Until the FCA releases the questionnaire and cross-tabs, the survey is evidence of use and mistaken expectations among a selected group, not a census of chatbot-led investing.
Four clear launches, no market total
Targeted Support took effect on 6 April. Under COBS 9B, an authorised firm can use limited information to align a customer with a predefined segment and deliver a ready-made suggestion designed for that group. It is less individualised than full advice, but it sits inside the regulatory perimeter and alongside the Consumer Duty.
The FCA estimates that about 23 million consumers are underserved by existing advice and guidance markets. Its latest count of nine permissions is progress, not proof that support has reached those people.
A review of firms’ public material finds at least four clearly launched consumer propositions. Royal London launched an app-based Targeted Support ISA service on 20 April. Quilter launched through Quilter Invest on 1 June. Vanguard now offers new clients an investment suggestion. The FCA says Monzo launched a Targeted Support journey earlier this year.
Other firms have announced permissions, planned roll-outs or use cases. Those announcements do not reveal active users. Nor has the FCA published a current list separating permission holders from live services, or aggregate figures for completed journeys, funded accounts, balances, complaints and investment outcomes.
Monzo’s early result illustrates the problem. The FCA said customers receiving an investment recommendation were 33% more likely to open an investment account than comparable customers who did not. That is a relative uplift. The 18 September speech did not disclose either group’s absolute conversion rate, sample size, selection method, balances, persistence, complaints or returns. A movement from 3% to 4% and one from 30% to 40% both produce roughly the same relative claim while describing very different reach.
A chatbot’s features do not settle its regulatory status
The FCA’s 2026 perimeter report says general-purpose large language models are outside its remit, while a model deployed specifically to provide financial advice would likely fall inside. The purpose and substance of the service matter more than whether its interface looks like a chat window.
Retrieving source documents or explaining jargon does not by itself create regulated advice. Adding personal prompts can make an answer more tailored, but is not a protection label. An affiliate link can bring financial-promotion rules into play if the surrounding words invite or induce investment activity, yet an approved promotion is still not the same thing as advice. A tool embedded by an authorised firm may be subject to that firm’s conduct obligations, but customers still need to know whether they are receiving information, Targeted Support or a personal recommendation, and whether the firm holds the relevant permission.
The protection boundary is concrete. A customer unhappy with regulated Targeted Support can complain to the firm and then take an unresolved case to the Financial Ombudsman. If an authorised provider fails, eligible Targeted Support claims may receive FSCS protection up to £85,000 per person, per firm. General-purpose chatbot output carries neither route merely because it cites sources, remembers preferences or links to a platform.
The next test is disclosure
The FCA said in its policy statement that it would assess firm numbers, customer uptake and suggested actions within two years of opening the gateway, with outcomes examined later. That timetable is too slow for basic launch transparency.
A quarterly dashboard could name permission holders, distinguish live from planned services and report completed journeys, funded accounts, opt-outs, complaints and broad outcome measures. Firms could also publish absolute conversion rates beside relative uplifts.
Britain has built a protected middle tier between generic information and full advice. It has not yet shown how much of the market that tier reaches. Without comparable denominators, the honest finding is neither that chatbots have won nor that Targeted Support has closed the advice gap. It is that one channel is being measured by a loosely defined survey response and the other by permissions.
Discussion
Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.