Using AI for money advice: UK, US and global data

How many UK consumers use AI for money

The FCA commissioned Yonder Consulting to survey 5,026 UK consumers with a day-to-day bank account, about 98% of UK adults, in April 2026. The findings were published with the FCA's Mills Review on 6 July 2026. The sample was weighted to be representative by age, gender, ethnicity, region, housing tenure and internet ability. The headline adoption figure is broad: 67% of consumers use AI for any purpose. Money is a smaller but real slice of that activity. 16% use AI for at least one personal finance task, rising to 23% of AI users.

What people ask the tools to do varies. Among those using AI for money tasks, 72% use it to summarise or explain information, 61% to suggest what to do, 24% upload personal documents such as bank statements, and 14% give it ongoing access to accounts or data. The survey also looked at specific shopping moments. Around a third of those shopping for pensions or investments used AI, and nearly half of the minority seeking debt advice did so. Most did not act on the first answer. Between 62% and 80% compared AI advice with other sources depending on the product, and 14% to 25% followed the AI's suggestion.

The protection most people don't know is missing

Only 40% of UK consumers correctly recognised that there is no formal route to put things right if something goes wrong after using a general-purpose AI tool such as ChatGPT or Gemini for investment advice. The share choosing a wrong answer, such as believing the Financial Ombudsman Service would apply, was 21% of all consumers, 24% of AI users and 41% of people who use AI for personal finance. The people closest to the technology were the most likely to hold the wrong belief about what stands behind it.

The reason is structural. The Financial Ombudsman Service handles complaints about businesses it covers, such as FCA-authorised firms, and the FSCS protects customers of authorised firms that fail. A general-purpose chatbot is not an authorised adviser, so neither applies to its suggestions. If an FCA-authorised firm uses AI in its own advice or service, the firm stays responsible and normal complaint routes apply. General information, not legal advice. Kael Tripton's reading, stated as analysis, is that using AI to understand a product is different from relying on it to choose one. For regulated decisions such as investments, pensions and debt solutions, the protection sits with authorised firms and the FCA register, not with the chatbot.

How much control people will hand over

Willingness falls as AI takes more control. 36% of UK consumers would use AI that reviews accounts and recommends, 30% would use AI that acts with permission each time, and 20% would use AI that acts autonomously within pre-set goals. Among AI users the figures are higher: 48%, 40% and 28%. The FCA equates 20% to about 11 million UK adults. Insurance reviews were the most attractive use, and AI moving money between accounts on its own the least.

Access to data follows a similar pattern. 26% see AI tools as a reliable source of financial information or advice. 13% of all consumers and 18% of AI users would give AI real-time access to their banking data. The gap between the 36% who accept recommendations and the 20% who accept autonomous action shows where the comfort line sits. Reviewing and suggesting is tolerated by more than a third. Acting alone within goals is tolerated by a fifth. The FCA's conversion of that fifth into roughly 11 million adults puts the scale of autonomous use in context, even though most of those people would still be relying on a tool that carries no complaint route.

What worries people most

The top concerns in the FCA survey were data misuse at 68%, being unprotected if something goes wrong at 67%, a few large companies gaining too much power over people's finances at 65%, being hacked at 63%, and AI recommending what benefits companies rather than the customer at 63%. The second of those, being unprotected, sits directly alongside the finding that only 40% know there is no formal route to complain. Concern about protection is high even where knowledge of the protection gap is low.

On what would change behaviour, 24% said nothing would persuade them. The top trigger was clear protection rules at 32%. That makes protection rules the single most cited condition for greater use, ahead of any product feature. The concern list also shows that the worries are not only about errors. Data misuse and corporate power rank near the top, which places the debate about AI in finance alongside broader questions about concentration and control, not just about whether a chatbot gives a wrong answer.

The US picture: fast adoption, low reliance

Pew Research Center surveyed 5,119 US adults on 17 to 23 February 2026. 44% said they use ChatGPT, up from 34% who had ever used it in Pew's February to March 2025 survey of 5,123 adults. Pew notes the question wording changed in 2026. About half of US adults say they never use a chatbot. Adoption of the tool is therefore broad and rising, but it is not universal, and a large block of adults remains outside it.

Use for money is much narrower than general use. In the Financial Health Network's Financial Health Pulse survey from April to June 2024, only 3% of US households used general AI chatbots such as ChatGPT for financial information, against 42% who turned to their bank or financial institution, the most-used source. The FINRA Investor Education Foundation's 2024 report found that very few consumers knowingly turn to AI for personal finance information, and that more consumers trust individual financial professionals than AI overall. The UK and US figures come from different surveys asking different questions in different years. They show direction, not a like-for-like comparison.

The industry side: firms are further ahead than customers

The Cambridge Centre for Alternative Finance's 2026 Global AI in Financial Services Report, published in May 2026, found 81% of surveyed financial services firms adopting AI at some level, with 40% of industry respondents at advanced stages. That is more than double the 20% of regulators at that level. The top three cloud providers serve more than 80% of industry respondents. The industry picture is one of widespread adoption, a substantial advanced cohort, and a regulatory community that is behind on the same measure.

Set that against the consumer picture and the three positions separate. Firms are adopting AI faster than their customers. UK consumers use AI widely at 67% for any purpose but only 16% for a personal finance task. US consumers use ChatGPT at 44% but only 3% of households used general AI chatbots for financial information in 2024. Regulators lag the industry on advanced adoption. The result is a market where the tools are being built into financial services faster than the public is using them for money, and faster than the rulebook and the complaint routes have adapted to the general-purpose tools people already have on their phones.

Using AI safely for money decisions

The practical line that follows from the data is between understanding and choosing. Using AI to summarise or explain information is the most common UK use at 72% of those using AI for money tasks. Using it to suggest what to do is also common at 61%. The protection question bites hardest at the point of choice, because a general-purpose chatbot is not an authorised adviser and neither the Financial Ombudsman Service nor the FSCS applies to its suggestions. Checking with providers and the FCA register, and keeping regulated decisions with authorised firms, is where the protection sits.

The survey evidence supports checking rather than blind following. Between 62% and 80% of UK consumers compared AI advice with other sources depending on the product, and only 14% to 25% followed the AI's suggestion. That behaviour is the safeguard most people already practise. The gap is knowledge of what happens when the check fails and the tool was a general-purpose chatbot rather than an authorised firm. Only 40% know there is no formal route to complain, and 41% of people who use AI for personal finance believe a wrong answer such as the Ombudsman applying. General information, not legal advice.

Source: FCA consumer research PDF.

Related coverage on Kael Tripton: How to Complain to the Financial Ombudsman Service: Step-by-Step Guide, Financial Ombudsman Service Insurance UK: How to Escalate a Complaint, FCA Register: How to Check If a Financial Firm Is Authorised, UK FCA Register: Checking a Financial Company, When can you access your pension? 55 now, 57 from 2028.

Frequently asked questions

Does the Financial Ombudsman Service cover advice from ChatGPT or Gemini?

No. The Financial Ombudsman Service handles complaints about businesses it covers, such as FCA-authorised firms. A general-purpose chatbot is not an authorised adviser, so the service does not apply to its suggestions. Only 40% of UK consumers correctly recognised this in the FCA's April 2026 survey.

Does the FSCS protect me if an AI chatbot gives bad financial advice?

No. The FSCS protects customers of authorised firms that fail. A general-purpose chatbot is not an authorised firm, so the scheme does not apply to its suggestions. If an FCA-authorised firm uses AI in its own advice or service, the firm stays responsible and normal complaint routes apply.

How many UK consumers use AI for personal finance tasks?

16% of UK consumers use AI for at least one personal finance task, rising to 23% of AI users, according to the FCA survey of 5,026 UK consumers in April 2026. 67% of consumers use AI for any purpose.

How many UK consumers would let AI act on their money autonomously?

20% of all consumers and 28% of AI users would use AI that acts autonomously within pre-set goals. The FCA equates 20% to about 11 million UK adults. 36% would use AI that reviews accounts and recommends, and 30% would use AI that acts with permission each time.

How many US adults use ChatGPT?

44% of US adults said they use ChatGPT in Pew Research Center's survey of 5,119 adults on 17 to 23 February 2026, up from 34% who had ever used it in Pew's February to March 2025 survey. Pew notes the question wording changed in 2026, and about half of US adults say they never use a chatbot.