Four times AI tried to lead my finances astray

Almost a quarter of UK investors use Artificial Intelligence (AI) to help them make investment decisions, according to new research from Fidelity International.¹ The study, which surveyed 13,000 investors around the world including 1,000 in the UK, found that young people were even more likely to rely on the technology.

Among 18- to 34-year-olds in the UK, more than a third (36%) use AI when making investment choices.

As a 30-year-old with an active interest in investing, I have been playing around with AI tools to see if and where they could improve my personal finances. Overall, I see some benefits - but with a few crucial words of warning.

Here I outline four times AI tried to lead my finances astray, along with my assessment of where I believe the technology can help us manage our money and where it falls flat.

• Need that next level of support? Fidelity’s financial advisers can help with savings, investments, retirement and pensions

1. Using old UK tax brackets

An AI assistant and I were having a chat about tax efficiency. It reminded me to watch out for additional rate tax, which starts at £150,000.

Additional rate tax is not something most of us need to worry about. Even so, the sentence made my warning lights flash.

I have written the phrase “Additional rate tax starts at £125,140” enough times to have that awkward number seared into my brain.

The threshold had been £150,000 but was cut back in 2023 - which means the AI’s warning was three years out of date. Someone earning £140,000 who read that statement might have been led to think they should only be paying tax at 40% rather than 45% - and later be painfully disabused of that misconception.

• Get the latest market updates, ISA news and pension insights by email

2. Claiming that policy changes are mere speculation

On another occasion, we were discussing the upcoming reduction in the cash ISA allowance, from up to £20,000 per year to £12,000.

The AI model warned me that this policy was not officially confirmed, and I should avoid acting based on speculation.

Generally, I would approve of such caution. However, the cash ISA cut has been confirmed by government for 2027, just the exact mechanics of enforcing it remain to be ironed out. If I did wish to make the most of the £20,000 allowance while it is still available, holding off thinking the news is mere “speculation” would not be my wisest move.

• Read: How ISAs could change in 2027
• Learn more about our award-winning Stocks and Shares ISA

Forgetting my time horizon

In a separate tête-à-tête, I asked: “I’m 30 - what should I be investing in?”

To give the AI assistant credit, it told me that, before investing aggressively, I should make sure I have:

• An emergency fund (typically 3–6 months of expenses)
• High-interest debt paid off (credit cards, etc.)
• Adequate insurance coverage

Once those were in place, it said I should consider a portfolio of 80-100% global stocks, or 60-80% stocks and 20-40% bonds if I wanted a smoother ride.

But it failed to ask if I should be investing at all. What if I was investing money that I wanted to use for a house deposit in two years? What if I was borrowing money to invest or using cash that I could well need in the short-term?

A golden rule of investing is never to put money into the stock market that you may need in the next few years. Ideally, you should be happy to leave that money for at least five years - although 10 or more is preferrable.

Its response was fine for me. But let’s imagine someone who didn’t know that golden rule asked the same question and then invested their house deposit fund that they plan to use in a year’s time, the consequences could be severe.

3. Ignoring my existing investment position

I decided to ask ChatGPT: “If I was going to buy one stock, what should it be?”

Ok, I confess. I was trying to catch it out by now and force it into giving me poor guidance with a loaded question.

Within seconds, it had pinged me a response - Microsoft shares - followed by a top five list of stocks for me to consider.

Admittedly, it did go on to say that, if my goal was simply to maximise the odds of long-term wealth building, I should choose a broad index fund over any single stock and asked me some questions about my age and how much I wanted to invest.

But any investment adviser worth their salt knows that, really, those questions (and that guidance) should be front and centre before any kind of single stock recommendation is given.

Also, the response completely overlooked my existing investment position. What if I already owned a significant proportion of Microsoft stock? Buying more could leave me perilously over-exposed. Ditto if I already had a portfolio significantly concentrated in the US technology sector or even just owned broad-based index funds which are heavily weighted towards the big tech companies.

The conclusion

I wouldn’t suggest AI models don’t have a place in building our financial plans. But these are four instances where following their guidance blindly could have led me astray. There have been countless other occasions where AI has proven very useful for my personal finances - from double checking some number crunching I’d done on the benefits of putting more money into my pension to providing a list of primary research I should read on the pros and cons on investing in emerging markets.

Personally, where I find AI most useful is when I have enough prior understanding of the subject to:

• Craft a robust prompt that is going to give me a balanced answer drawing only on reliable sources
• Think critically about the response and know where it is lacking or plain wrong
• Find the primary sources to double check its answer

Think of it like a well-meaning but sloppy assistant: eager to please you but potentially happy to take shortcuts along the way and often unable to see the bigger picture of what you need.

Sure, you’d trust them with your diary management, but would you trust them managing your pension?

Got a burning question you want to ask? Why not drop us a line. Click here to ask your question.

• Read: Could the AI boom be masking economic fragility?
• Read: Why physical AI is the next investment frontier
• Read: Top 10 best-selling ISA and SIPP funds in May

Sources:

¹ Fidelity International Be Invested 2026 Study. The survey of 13,000 retail investors was conducted 12 February - 11 March 2026 by Opinium. Countries in scope included Germany (1,000), France (1,000), Italy (1,000), Netherlands (1,000), Spain (1,000), Switzerland (500), UK (1,000), Hong Kong (1,000), Singapore (1,000), Taiwan (1,000), Australia (1,000), Japan (1,000) & China (1,500)

Important information: investors should note that the views expressed may no longer be current and may have already been acted upon. Reference to specific securities should not be construed as a recommendation to buy or sell these securities and is included for the purposes of illustration only. Tax treatment depends on individual circumstances and all tax rules may change in the future. There is a risk that the issuers of bonds may not be able to repay the money they have borrowed or make interest payments. When interest rates rise, bonds may fall in value. Rising interest rates may cause the value of your investment to fall. This information is not a personal recommendation for any particular investment. If you are unsure about the suitability of an investment you should speak to one of Fidelity’s advisers or an authorised financial adviser of your choice.