AI Knows a Lot About Money - Just Not Yours
- Chris Howell
- 2 days ago
- 7 min read
AI assistants can explain mortgages, pensions, tax, investments and business finances in remarkably clear language.
That does not mean they understand your financial circumstances.
Tools such as ChatGPT, Claude, Google Gemini, Microsoft Copilot and Perplexity can be useful starting points when you are trying to make sense of money. They can translate jargon, organise information and help you prepare questions.
The danger comes when a helpful explanation begins to feel like personalised financial advice.
Research from the UK has found that AI assistants can produce incorrect figures, overlook important details and present uncertain answers with complete confidence. Unlike advice from an authorised professional, their answers normally come with no protection from the Financial Ombudsman Service or Financial Services Compensation Scheme.
So, where can AI genuinely help—and where should you stop listening?
AI can be useful for understanding money
There is nothing inherently reckless about asking an AI assistant a financial question.
Used carefully, AI can help you:
Explain unfamiliar financial terms in plain English
Create a basic household or business budget
Compare generic financial scenarios
Summarise a mortgage offer, pension statement or insurance document
Organise receipts and suggest expense categories
Prepare questions for an accountant or financial adviser
Build a spreadsheet for tracking spending or cash flow
For example, you might ask:
What is the difference between a fixed-rate and tracker mortgage?
That is a general information question. An AI assistant can usually explain the broad differences, including how interest rates affect payments.
You could also ask it to produce a list of questions to discuss with a mortgage adviser. That may help you arrive at the appointment better prepared.
Similarly, AI can explain terms such as annual percentage rate, compound interest, tax allowance or pension drawdown without telling you what you personally should do.
These are lower-risk uses because the AI is helping you understand or organise information rather than making the final decision. The research report found that generic explanations, document summaries, budgeting templates and adviser preparation were among the more appropriate uses, provided the output was checked.
The problem is that AI can sound right when it is wrong
AI assistants are very good at producing confident, fluent answers.
Unfortunately, confidence and accuracy are not the same thing.
A 2025 Which? investigation tested six popular AI services using 40 questions across consumer finance, law and health. Reported accuracy ranged from around 51% to 71%, depending on the tool. ChatGPT scored 64% in that test.
Some errors were not minor. Researchers found examples of AI systems:
Supporting actions that would breach the annual ISA allowance
Giving incorrect information about travel insurance
Misstating flight-delay compensation rules
Suggesting a payment-withholding approach that could create contractual problems
Directing users towards paid tax-refund services without clearly highlighting the free HMRC route
The assistants also regularly failed to account for differences between financial and legal rules in England, Scotland, Wales and Northern Ireland.
This matters because AI does not necessarily recognise that your question contains a false assumption.
It may answer the question exactly as asked rather than stopping to say:
The figure in your question appears to be wrong.
That makes financial prompts particularly risky. A plausible-looking calculation built on the wrong allowance, tax rate or interest assumption can still produce a very polished answer.
AI does not know your full financial situation
The title of this article is the heart of the issue.
AI knows a great deal about money in general. It does not know your complete financial life.
It may not know:
Your income and outgoings
Your existing debts
Your attitude to risk
Your pension arrangements
Your dependants
Your employment status
Your tax position
Your health
Your future plans
Which part of the UK you live in
Whether the information you supplied is complete or accurate
Even when you provide some of those details, the AI cannot independently verify them.
A regulated adviser has professional responsibilities. They are expected to gather relevant information, understand your circumstances and explain why a recommendation is suitable.
A general-purpose chatbot has none of those obligations.
It generates a response based on your prompt, its training and any information it can retrieve. It does not truly know whether the answer is right for you.
Financial information is not the same as financial advice
This distinction is important.
Financial information
This means general facts or explanations, such as:
What does APR mean?
or:
How does compound interest work?
This information is not tailored to your personal circumstances.
Financial guidance
Guidance can help you understand your options without making a specific recommendation.
MoneyHelper, for example, provides free and impartial financial guidance. It can help people understand money problems and decide what kind of support they may need.
Regulated financial advice
Regulated advice involves a recommendation based on your circumstances, often involving a specific financial product or course of action.
Examples might include recommending:
A particular pension arrangement
A specific investment
A mortgage product
An insurance policy
A debt solution
In the UK, regulated financial advice must be provided by an appropriately authorised firm or individual.
General AI tools are not regulated financial advisers. The FCA has also made clear that information from general-purpose AI systems does not receive the same complaint and compensation protections as regulated advice.
When you should not rely on AI alone
The higher the financial stakes, the less appropriate it becomes to rely on an AI-generated answer.
You should not use an AI assistant as the sole basis for decisions involving:
Investments
AI can explain different types of investments and describe general risks.
It should not decide which investment is suitable for you, how much you should invest or whether you can afford a potential loss.
Pensions
Pension decisions can have long-term and sometimes irreversible consequences.
AI may help explain terms in a statement, but pension transfers, withdrawals and drawdown decisions require personalised consideration.
Debt solutions
Individual voluntary arrangements, bankruptcy, debt management plans and other debt solutions have serious legal and financial consequences. Use an appropriate debt-advice organisation or qualified professional rather than relying on a chatbot’s recommendation.
Tax returns
AI can explain general tax concepts, but tax allowances, rules and deadlines change.
An incorrect answer can lead to overpayments, missed allowances, penalties or compliance problems.
Research involving UK accountants found that half were aware of clients suffering direct losses following incorrect AI-generated tax or financial guidance. Reported problems included misclassified expenses, incorrect VAT treatment and flawed tax planning.
Mortgages and insurance
AI can explain general product differences, but it cannot reliably determine suitability based on your complete circumstances. The final choice should be checked through an appropriate regulated or qualified source.
Be careful what financial information you upload
Accuracy is not the only risk.
People often provide AI tools with far more information than they realise, particularly when asking for help with tax, budgeting or business accounts.
Avoid pasting sensitive information such as:
Bank-account numbers
National Insurance numbers
Complete tax records
Payroll details
Customer financial information
Client invoices containing personal data
Pension account details
Unredacted business accounts
This is especially important for freelancers, sole traders, charities and small organisations handling information about other people.
UK data-protection responsibilities still apply when personal information is processed using generative AI. Businesses may need a lawful basis, suitable contractual arrangements and a clear understanding of how the provider stores or uses submitted data.
Business and enterprise AI accounts can offer stronger controls than free consumer accounts. These may include agreements preventing customer data from being used for model training, administrative controls, audit logs and configurable data retention.
However, a business subscription is mainly a privacy and governance improvement. It does not automatically make the financial answer more accurate.
Seven ways to use AI more safely for money questions
AI can still be useful, but it needs supervision.
1. Use it to understand, not decide
Ask AI to explain terminology, outline options or help you prepare questions.
Do not ask it to make the final decision for you.
2. Check every important figure
Verify tax rates, allowances, pension thresholds, benefits and regulatory rules using official sources such as GOV.UK, HMRC, the FCA or MoneyHelper.
3. Check the date
Financial rules change. An answer that was correct last year may be wrong today.
4. Ask for assumptions
Tell the AI to list every assumption used in a calculation or comparison.
This makes it easier to spot missing information or incorrect starting figures.
5. Follow the sources
A citation is useful only when it genuinely supports the claim. Open important links and check what the original source says.
6. Protect sensitive data
Remove names, account details and identifiable information before using AI to analyse a document. For business or client information, use an appropriate business-grade system and check the provider’s data terms.
7. Know when to involve a professional
Use AI to prepare for a conversation with an accountant, adviser or money-guidance service—not to avoid that conversation when the decision is important.

A useful assistant, not a financial adviser
AI can make financial information easier to understand. That is valuable.
A person who previously found pensions, mortgages or tax completely impenetrable may use AI to learn the basic language, organise their thoughts and ask better questions.
But the same friendly, confident tone can make an incorrect answer feel trustworthy.
The safest approach is to treat AI as a research and organisation assistant. Let it explain, summarise and help you prepare.
Do not let it make significant financial decisions on your behalf.
After all, AI may know a lot about money.
It still does not know about yours.
Using AI in your organisation?
Mercia AI’s AI Ethics and Responsibility Workshop helps organisations understand the risks, responsibilities and practical safeguards involved in using AI at work. This includes data handling, human oversight and knowing which decisions should not be delegated to automated tools.
This article provides general information about using AI and is not financial, tax, legal, investment or debt advice.


