AI ISA advice: I tested four tools on the questions people get wrong

I run these tests so I know what AI ISA advice I can trust. ISA rules looked like a good thing to push on. They’re UK-only, they’re specific, and they change on a known date (6 April is when the new tax year and most rule changes land). That last part matters: a fixed reform date splits AI tools cleanly by how current their knowledge is. An ISA, for anyone arriving from the AI side of this site, is a UK tax-free savings account: you can pay in up to £20,000 a year and keep the gains tax-free.

So I started with the basics. Total allowance, holding more than one ISA of the same type, the Lifetime ISA limit, flexible ISAs, tax inside the wrapper. Four tools, same questions. Every one came back correct.

Four AI tools sat the same HMRC quiz and all four passed without dropping a mark, not the result I was hoping for from a test designed to catch something out.

So I moved to the questions people get wrong.

The basics: nobody put a foot wrong

I put five standard questions to ChatGPT (Free), Claude (Opus 4.8), Gemini (Pro) and Perplexity (“Best”), and watched the same correct answers come back four times over. How much you can put in across all ISA types in a year (£20,000). Whether you can now hold more than one cash ISA in the same year (yes, since April 2024). The Lifetime ISA limit and its government bonus (£4,000 a year, 25% on top). How a flexible ISA lets you take money out and put it back without losing allowance. Whether you pay tax on gains inside an ISA (you don’t).

All four got all five right. Three of them named the April 2024 change to the multiple-ISA rule without being asked. That tracks with what these tools are built on: ISA basics are documented everywhere (gov.uk, every bank, every comparison site) and they don’t move much. Stable, well-indexed facts are exactly what a model recalls well. This is the part of the test that matters most for what comes next, because it rules out the lazy conclusion. The tools are not generally bad at ISA rules. They’re specifically shaky in one place.

This is the failure mode I’ve named elsewhere as stale data, fresh confidence: a model reporting a rule from the last time it was stable, with no sign it knows the rule has moved on. The basics don’t trigger it, because nothing here has changed enough for the old answer to be wrong.

The harder questions

The second round was four questions that trip up real people: the kind you’d actually type into an AI before making a decision, not contrived puzzles. I sent the same four to each tool, framed as a UK resident in the 2025/26 tax year:

1. The shared allowance. I’ve paid £15,000 into a cash ISA this year. Can I add £10,000 to a stocks and shares ISA too? (The £20,000 limit is shared across all ISA types, so the answer is no: only £5,000 left.)
2. An inherited allowance. My husband died holding about £60,000 across his ISAs. Does my own allowance change? (Yes, a one-off extra allowance called an Additional Permitted Subscription, or APS, equal to his ISA holdings, on top of my normal £20,000.)
3. A partial transfer. I’ve paid into my stocks and shares ISA this year and want to move it to a cheaper provider. Can I transfer just part of it, or must I move all of this year’s money? (This is the one. More below.)
4. An age limit. I’m 17. Can I open an adult stocks and shares ISA? (No, adult ISAs are 18 and over.)

Questions 1, 2 and 4 are stable rules, well documented, no recent change. Question 3 is not. The rule on partial transfers of current-year money changed on 6 April 2024, and that change is newer than a lot of what’s been written about ISAs online. If a tool is going to get caught out, this is where.

The scoreboard

I ran all four the same way: same four questions, same framing, UK resident in the 2025/26 tax year. Here’s what came back.

Three clean rows. One split. The split is the interesting one.

Question 3: where two tools gave a rule that died in 2024

Before 6 April 2024, if you wanted to move ISA money you’d paid in during the current tax year, you had to move all of it: the whole year’s contributions, or nothing. The April 2024 reforms scrapped that. Since then you can transfer part of this year’s money and leave the rest where it is. The change is on gov.uk’s transfer guidance, and M&G, MoneyHelper and Royal London all confirm the same 6 April 2024 date.

When I asked, Perplexity gave the old rule. Verbatim:

“You can usually transfer just part of ISA money from previous tax years, but current-year subscriptions must be transferred in full if you transfer them at all… So if the money you want to move was paid into your stocks and shares ISA this tax year, it must go in full.”

ChatGPT said the same thing:

“But current tax year subscriptions must be transferred in full for that ISA account… If the money you want to move includes this tax year’s contributions, you must transfer all of this year’s subscriptions from that ISA (not just a slice of them).”

Neither attached a date. Neither hedged. ChatGPT answered from memory; Perplexity searched the web and cited its sources, and gave the dead rule anyway.

Both recited a rule HMRC had buried two years earlier, with the straight face of a tool reading confidently from a rulebook two editions out of date.

(ChatGPT was running on a reduced model at the time. A free-tier notice appeared mid-answer saying it was “using a less powerful model until your limit resets.” The error was the same; it isn’t a tier story.)

Here’s why this one matters more than a wrong allowance figure would. A reader who trusts that answer believes they’re locked in. If they wanted to move to a cheaper provider and only shift part of this year’s money, they’d conclude they can’t, and either stay put on the higher fees, or move the whole lot when they only meant to move some. The wrong answer doesn’t just misinform; it removes an option that’s open to them.

Now the other two. Claude opened its response, before answering anything, with this:

“I’ll verify the current rules before answering precisely, since ISA rules have seen recent changes and reform discussion.”

It then ran four web searches, cited OakNorth, M&G, Unbiased, Shawbrook and MoneyHelper, gave the correct post-2024 answer, and named the April 2024 change explicitly. It also volunteered, unprompted, that the cash ISA limit is being cut to £12,000 for under-65s from April 2027, a change to a question I hadn’t asked.

Gemini got there too, and said the quiet part out loud:

“Following a government rule change in April 2024, you are now allowed to make partial transfers of current-year ISA contributions. Prior to this, you were forced to move current-year money in its entirety.”

That’s the whole finding in one sentence. The rule changed; the answer changed with it. Gemini cited Moneyfacts and Magee Gammon inline.

The pattern is almost clean, and the exception is the interesting part. Claude and Gemini searched before answering and got the changed rule right. ChatGPT answered from memory and gave the rule from the last time it was stable. Perplexity searched too, cited its sources, and still gave the dead rule: proof that searching isn’t the whole story, it’s what the tool does with what it finds. None of the four was a worse tool than the others on the basics. They all aced those. It’s the same lesson I keep finding when I test these tools side by side on a real question: a tool that searches usually beats one that answers confidently from memory, but a search is only as good as the source it lands on.

The tell, and the one prompt that catches it

What I keep coming back to is that the tell was sitting in plain view: no date attached, and no hedge. Perplexity even searched and cited sources, and still didn’t date its answer, which is the point: the missing date is the tell, not the missing search. When an AI answers a question about something that changes on fixed dates (tax rules, ISA limits, benefit thresholds) and gives you a flat answer with no date attached, the answer might be from whenever that rule was last settled, not from now.

The fix is one line you add to the question. Make the tool cite the rule and the date the rule took effect:

That’s not a guarantee. But it makes the tool work harder to confirm it’s current, and it turns a vague claim into a testable one. “Under ISA rules, current-year money must move in full” can’t be checked. “Under the rules as of [date], you can move part of it. This changed on 6 April 2024” can. It’s also, when you look at it, just asking the AI to do the one thing that would have stopped the error: tell you when the rule it’s quoting actually applied. There’s a version of this idea in the Prompt Stack, the step where you separate what’s an established fact from what the model is assuming.

One more thing worth knowing if you act on a correct answer here: the law allowing partial transfers doesn’t force your provider to offer them. Both Claude and Gemini flagged this. The rule permits it; your provider still has to support it on both ends. Check with both before you assume.

If you only ever ask one type of AI question about money, a free tool that searches the web is the one to use, and check it actually searched on the question that matters, not just the easy ones around it.

Caveats

This is a small test, and I’d rather you knew its limits than took it as a league table. Three things to hold:

Claude wasn’t a clean-room run. Claude.ai has no memory-off mode, and my account has prior ISA conversations in it. Some of Claude’s strong showing here may reflect that context rather than the model alone. I can’t rule it out, so I won’t claim Claude is simply “more careful” on the strength of one run.

It’s one run per tool. N=1 in each cell, not a hit rate. The Question 3 miss is real and verified against gov.uk, so it stands as a dated, point-in-time result, but models get updated, and this specific failure may not reproduce next month. The lesson outlives the result: ask for the date.

ChatGPT was on a reduced model for part of its answer, as noted above. It doesn’t change the finding, but it’s on the record.

The honest read isn’t “don’t trust AI on your ISA.” Three of four questions were clean across every tool, and even the two that missed Question 3 would have been right a year earlier. The read is narrower and more useful: trust it least on the things that have recently changed, and the cheapest way to find out whether something changed is to make the tool tell you the date.