City AM reports that the FCA has warned consumers using AI for investment decisions that they may have little or no regulatory protection if things go wrong. That is an important point. But it also raises a bigger question that the warning itself does not really answer: what is the realistic alternative for people who cannot access, afford or justify traditional financial advice? If the choice for many people is not “AI or a financial adviser”, but “AI or no meaningful help at all”, then the correct benchmark is not perfection. It is the outcome available to the same person without AI.
A curious warning appeared in the financial press this week.
“Use AI for investing at your own risk, warns watchdog.”
The message is straightforward.
More people are turning to tools such as ChatGPT and Gemini for help with investing. General-purpose AI is not regulated by the Financial Conduct Authority. If it gets something wrong, you cannot complain to the Financial Ombudsman or claim compensation from the Financial Services Compensation Scheme simply because the chatbot gave you a bad answer.
That is important information.
But it is not the whole story.
Because once you look beyond the warning, a rather different question appears:
What are people supposed to use instead?
The comparison is AI versus an adviser
The reality is often AI versus nothing
The City AM article reports that four in five less-experienced investors have used AI for help with investing.
That figure comes from FCA research involving 666 UK respondents aged 18–40 who either owned investments or were considering investing. Fifty-six per cent said they trusted AI tools, while 44% incorrectly believed AI-generated financial information was regulated.
Those are useful findings.
But there is another FCA statistic that changes their meaning considerably.
Only around 9% of UK consumers take regulated financial advice.
So around nine people in ten are not currently taking regulated advice.
That does not necessarily mean that every one of them is unable to obtain it. Some do not want it. Some do not need it.
But affordability and accessibility are sufficiently significant problems that the FCA itself describes an “advice gap” and says many people struggle to access affordable financial advice.
That produces a very different decision tree.
For millions of people, the realistic alternatives may not be:
AI or a highly qualified financial adviser?
They may be:
AI or Google.
AI or TikTok.
AI or a friend.
AI or guessing.
AI or doing nothing.
And that comparison matters enormously.
AI does many of the things advisers say it cannot do
One adviser quoted in the article says AI cannot necessarily replicate the personalised assessment required to understand someone’s:
objectives, time horizon, appetite for risk and capacity for loss.
But why couldn’t it?
Those are not mysterious properties accessible only through human intuition.
An AI can ask:
What are you trying to achieve?
When will you need the money?
What other assets do you have?
What liabilities do you have?
How secure is your income?
How would a 30% fall affect your plans?
How would you feel about it?
What would cause you to sell?
What assumptions are you making?
What have you forgotten to consider?
Indeed, AI can keep asking questions until the user understands the question.
It can explain terminology at five different levels of complexity.
It can identify contradictions.
It can model alternative scenarios.
It can challenge assumptions.
It can compare options.
It can remember information supplied during the conversation.
It can ask the user whether the conclusion still feels right after seeing the consequences.
None of this proves that AI is infallible.
It plainly isn’t.
But the FCA itself acknowledges that AI can summarise complicated subjects, make information easier to understand and make investment research more efficient.
So the interesting question isn’t whether AI can make mistakes.
Of course it can.
The question is:
Compared with what?
Where is the comparative risk assessment?
This is perhaps the biggest omission in the story.
We are told that AI can be wrong.
But we aren’t told how often.
We aren’t told how serious those errors are.
We aren’t told whether users detect them.
And, most importantly, we aren’t given any comparison with the alternatives.
Human advisers make mistakes.
Investment managers make mistakes.
Regulated firms fail.
Consumers misunderstand advisers.
People misunderstand documents produced by regulated firms.
Recommendations later turn out badly.
Entire compensation systems such as the Financial Ombudsman Service and FSCS exist precisely because regulation does not eliminate error or harm.
That does not make regulated advice unsafe.
It simply means that:
“AI sometimes gets things wrong” is not, by itself, a comparative risk assessment.
To establish that consumers are worse off using AI, we would need to know what would otherwise have happened.
That counterfactual is almost completely absent.
Protection and capability are not the same thing
This points to a deeper conceptual problem.
Financial services regulation understandably concentrates on protection.
Who is liable?
Who can make a recommendation?
Who is authorised?
Where can someone complain?
Who pays compensation if something goes wrong?
These are important questions.
But they are not the same question as:
What makes someone more capable of making a good decision?
A regulated adviser can give someone greater protection while leaving most of the thinking with the adviser.
AI can potentially give someone less regulatory protection while substantially increasing their own capacity to investigate, question, compare and decide.
That distinction matters.
At the Academy of Life Planning, we might call it the difference between:
protection from decisions and capability for decisions.
A healthy financial system needs both.
The BIG Checker spotted the same missing frame
I ran the article through The BIG Checker, our narrative-analysis tool.
It rated the article as presenting a medium narrative risk.
Its central finding was not that the FCA’s warning was false. Quite the opposite: it concluded that the article was accurate about the absence of regulatory and compensation protection.
The concern was what happened next.
The report found that industry voices were allowed to dominate the solution framing, moving the reader towards paid professional advice without properly acknowledging the affordability gap that helped drive consumers towards AI in the first place.
It identified a particularly important assumption being normalised:
professional financial advice becomes the benchmark against which AI is judged.
Meanwhile, the advice gap is treated almost as a fact of nature rather than something produced by the economics and architecture of the existing market.
And the omissions are revealing.
The article does not examine the cost of advice, clearly distinguish AI-assisted research from surrendering a decision to AI, compare outcomes with human advice, or explore the full range of alternatives available to consumers.
That creates what the report calls a false binary: an unsafe free option versus an inaccessible paid one.
There is another possibility.
Help people become better decision-makers.
Follow the incentives — but carefully
There is also an institutional question worth asking.
The FCA is independent of government and recovers its annual running costs through fees charged to authorised firms.
For 2026/27, the regulatory fee block covering advisers, arrangers, dealers and brokers has an annual funding requirement of £107.2m.
That does not prove that the FCA wants consumers to employ financial advisers.
Nor should it be presented as evidence of regulatory impropriety.
But systems thinking asks us to examine incentives even where everyone involved is acting conscientiously.
The regulated financial system naturally views the world through the regulated financial system.
It asks:
How do we bring more consumers safely into regulated support?
A consumer-agency perspective asks a slightly different question:
How do we give more people the capability to make good decisions, whether or not they subsequently employ someone?
Those questions overlap.
They are not identical.
Something even more interesting is happening
The FCA’s own recent work actually recognises much of this.
Its 2026 wealth-management research says AI could “reduce friction, improve efficiency and help close the advice gap”.
And the FCA’s Mills Review says AI could fundamentally reshape consumer journeys, competition and market power in financial services. It reports substantial consumer appetite for AI operating within personal finance.
So perhaps the real story isn’t:
Consumers are recklessly replacing advisers with chatbots.
Perhaps it is:
Consumers have discovered an intelligence layer that was previously unavailable to them.
That is much more disruptive.
For the first time, someone without £100,000 or £500,000 to invest can sit down with something capable of discussing diversification, sequence risk, asset allocation, taxation, inflation, pensions, behavioural bias and financial trade-offs for as long as they like.
They can say:
“I don’t understand.”
And ask again.
And again.
There is no clock running.
No embarrassment.
No minimum portfolio.
No product waiting at the end of the conversation.
That does not make AI a financial adviser.
It makes something else possible.
Intelligence is becoming abundant
Financial expertise has historically been scarce.
Scarcity created intermediaries.
Intermediaries became institutions.
Institutions created regulatory structures.
And eventually we came to assume that sophisticated financial thinking had to be delivered through those institutions.
AI challenges that assumption.
Intelligence is becoming cheap.
Explanation is becoming abundant.
Analysis is becoming available on demand.
The scarce resource increasingly isn’t information.
It is judgement.
And judgement need not mean surrendering the decision to either a chatbot or an adviser.
It can mean developing the capacity to use both intelligently.
That is the agency opportunity.
Don’t ask AI to take your agency
Use it to increase your agency
There is one part of the FCA’s warning with which I strongly agree.
The final judgement should remain yours.
The FCA’s Lucy Castledine puts it well:
AI can help people research companies, understand jargon and explore options, but consumers should continue to use their own judgement.
Precisely.
But I would go further.
Don’t simply ask:
“Can AI give me financial advice?”
Ask:
“Can AI help me become sufficiently informed to make a better financial decision?”
That is a very different proposition.
Ask it to explain.
Ask it for the strongest argument against its conclusion.
Ask it what information is missing.
Ask it to identify assumptions.
Ask it to model what happens if those assumptions are wrong.
Ask it what questions a competent financial planner would ask.
Ask it what a product salesperson might fail to mention.
Ask it to separate facts from judgement.
Ask it to show its sources.
Then decide.
This is not outsourcing judgement to artificial intelligence.
It is augmenting human intelligence.
The real consumer-protection opportunity
The financial services industry may therefore be asking the wrong question.
It keeps asking:
How do we protect people from unregulated AI?
There is a bigger question:
How do we teach people to use AI well enough that it increases their agency rather than replacing it?
Because 91% of consumers are not currently taking regulated financial advice.
Warning them that AI isn’t an adviser doesn’t solve that problem.
Making advisers slightly faster doesn’t solve it either.
AI presents another possibility.
Give millions of people access to understanding that was previously scarce.
Teach them how to interrogate it.
Teach them to verify important facts.
Teach them when uncertainty becomes significant enough to bring in specialist human expertise.
And let the human expert enter where the human expert adds value — rather than assuming that every financial question requires an ongoing advisory relationship.
Perhaps the future isn’t:
AI instead of advisers.
Nor is it:
advisers protecting us from AI.
It is something more empowering:
continuous agency, AI-assisted understanding and episodic human expertise.
And perhaps that is the safety net we should be trying to build.
Academy of Life Planning 