Most survey findings about artificial intelligence tell you more about the people answering than about the technology. The set the Financial Conduct Authority put out on 27 August 2026 is an exception, because one of the numbers in it is not an opinion. It is a factual error about who pays when something goes wrong.
Per the regulator’s press release, around a third of respondents — 32% — wrongly think they would get compensation from the Financial Services Compensation Scheme or the Financial Ombudsman Service if AI advice went wrong. Almost half, 44%, in the FCA’s phrasing “mistakenly believe AI-generated financial information is regulated”.
The methodology, stated up front
The FCA says the work was run through the platform Attest as a quantitative usage and attitudes study, with fieldwork on a single day, 24 July 2026, and a sample of 666 UK respondents aged 18 to 40 who own investments or would consider buying within twelve months. A modest base and one day in the field: a signal rather than a census. The regulator publishes the method alongside the numbers, which is the minimum a reader needs to weigh them.
Why a fintech desk is writing this up
Because the mistake is not about AI. It is the perimeter mistake, in a new costume.
We have watched it three times this month. A firm joining a regulator’s support programme reads to consumers as a seal of approval it is not. A register entry showing a temporary permission reads as full authorisation. Now a chatbot’s output reads as something with a compensation scheme behind it. Each time, the consumer has performed a check, got a result, and drawn a conclusion the result does not support.
The direction of the error is consistent: people assume the safety net is wider than it is, and that it follows whatever they are looking at rather than the firm they are dealing with.
What the regulator actually says
Lucy Castledine, the FCA’s director of consumer investments, is quoted in the release saying that AI can help you research companies or understand jargon, but that you need to understand how you are protected and continue to use your own judgement. That is a boundary statement, not a warning about the technology, and it is the part worth carrying into every other corner of consumer finance.
The reader’s version
Work out which named legal entity you are dealing with, and check that entity on the FCA Register rather than checking the brand, the app or the answer on your screen. Then read what the firm itself publishes about its complaints route. Neither step is exciting, and both bear on where you stand if the thing goes wrong.
A note on standing, since standing is the subject: Fin Watch is a publisher, not an authorised firm. We hold no Financial Conduct Authority permissions, we vouch for nobody else’s, and none of the above is financial advice.