Key Takeaways
- Millions of people are using AI chatbots for money advice — but most don’t realise their conversations may be stored or used for training
- Never type in account numbers, exact balances, loan reference numbers, or ID details into a public AI tool
- You can get genuinely useful financial guidance from AI without sharing a single piece of real personal data — by using placeholders and concept-based questions
- Most major AI platforms have a setting to stop your data being used for model training — and most people have never checked it
- For big decisions — property, retirement, major debt — a certified financial planner is still worth it
I came across a piece in The Week this week about what not to share when using AI for personal finance help — and honestly, I had to stop mid-scroll. Because I realised I’d probably done half the things they were warning about. And I’d bet most people reading this have too. The question of what not to share with AI for personal finance is suddenly everywhere, and for very good reason.
So I spent a few hours digging into the actual privacy policies, the data storage practices, and the specific risks. Here’s what I found — broken down so it actually makes sense.
Why Everyone Is Suddenly Using AI for Money Questions
Let’s start with the obvious: AI chatbots are incredibly useful for financial questions. You can ask them things you’d be embarrassed to Google, they explain complex concepts without the condescending tone of a bank brochure, and they’re available at 2am when you’re spiralling about your credit card balance.
A 2025 survey by NerdWallet found that nearly 40% of people under 40 had used an AI tool for some form of financial guidance in the past six months. That number has almost certainly grown since. Tools like ChatGPT, Gemini, Copilot, and a dozen fintech-specific AI assistants are now built directly into banking apps across Europe and Asia.
The problem isn’t the tools themselves. The problem is how people are using them — specifically, what they’re typing in.
What Not to Share With AI for Personal Finance (The Specific List)
Here’s where I want to be precise, because vague advice like “don’t share sensitive information” is useless. Let me tell you exactly what the risk categories are.
Account numbers and balances. This is the obvious one — but people do it constantly. Typing “my HSBC account ending in 4821 has €6,200 in it” into a chatbot is genuinely risky. Even if the AI itself is secure, that conversation exists somewhere on a server. If that server is ever breached — and breaches happen regularly — that data could end up in a leak.
Your exact income figure tied to your name or location. Asking “I earn €48,000 a year in Dublin, how should I invest?” sounds harmless. But that combination — specific income, location, potentially your email if you’re logged in — is the exact kind of profile data that gets harvested by data brokers. I’m not saying it’s happening with every platform. I’m saying you have no way to confirm it isn’t.
National ID numbers, tax reference numbers, or passport details. This should feel obvious but people paste these in when asking about tax filings or government benefit applications. Never do this. Ever.
Loan or mortgage reference numbers. These are unique identifiers. Combined with other data points, they can potentially be used for account takeover attempts if the data gets out.
“The safest mental model: treat an AI chat window like a postcard, not a sealed letter. Write only what you’d be comfortable with a stranger reading.” — data security researcher quoted in The Week, June 2026
What You CAN Share — And How to Reframe Your Questions
Here’s the thing that took me a while to understand: you can get genuinely excellent financial guidance from AI without sharing a single real number. The trick is to ask concept-based questions and use placeholders.
Instead of: “I have €23,000 in debt across two credit cards at 19% APR — should I do a balance transfer?”
Try: “If someone had €X in debt across two credit cards at around 19% APR — that’s the yearly interest rate — how would you weigh up a balance transfer versus paying down the highest-rate card first?”
The AI gives you the exact same quality of answer. And you’ve shared nothing identifiable.
Other genuinely safe questions:
- How does compounding interest actually work?
- What’s a realistic emergency fund size for a single person with no dependents?
- What should I look for in a low-cost index fund?
- What questions should I ask a financial advisor before hiring them?
- How do I read a credit report without any of the numbers I’m looking at right now?
Notice the pattern. You’re asking the AI to teach you, not to analyse your specific financial life. That distinction matters enormously — both for privacy and, honestly, for the quality of advice you get. AI isn’t a licensed financial planner. It can explain how balance transfers work. It cannot actually know whether one is right for your specific situation without knowing your full financial picture — and you shouldn’t give it that.
- Ihr Gesundheitsinfluencer hat wahrscheinlich keinen medizinischen Abschluss – das hat Pew Research gerade herausgefunden
- Imitações de Destinos Estão Cortando Orçamentos de Viagem pela Metade — E a Maioria dos Viajantes Ainda Está Pagando Preço Cheio
- Die Best Buy Memorial Day Sale 2026 ist jetzt live – Hier sind die Angebote, die die meisten Käufer vor Montag verpassen werden
The Setting Most People Have Never Checked
This part I genuinely didn’t know until this week. Most major AI platforms — including ChatGPT, Gemini, and Microsoft Copilot — have a setting that controls whether your conversations are used to train their models. By default, on many platforms, this is switched on.
For ChatGPT: Settings → Data Controls → toggle off “Improve the model for everyone.”
For Google Gemini: Go to myactivity.google.com → Gemini Apps Activity → turn off.
For Microsoft Copilot: The controls depend on whether you’re using a personal or enterprise account. Enterprise accounts often have stronger defaults, but personal users should check under Privacy settings.
Does turning this off guarantee your data disappears? No. But it significantly reduces how your conversations are processed and used beyond your immediate session. It takes about 30 seconds and most people have never done it.
| What You Share | Risk Level | Better Alternative |
|---|---|---|
| Exact account balance + bank name | 🔴 High | Use “around €X” with no bank name |
| Exact salary + city + age | 🟡 Medium | Income range only, no location |
| National ID or tax number | 🔴 High | Never. Use your government’s official portal instead |
| General financial concept questions | 🟢 Safe | This is exactly what AI is great for |
| Hypothetical scenarios with placeholder numbers | 🟢 Safe | Best way to get personalised-feeling advice safely |
What’s Your AI Finance Habit?
Answer 3 quick questions and get a personalized plan for using AI tools safely with your finances.
1. Have you ever typed your income or salary into an AI chat tool?
When AI Is Not Enough — And What to Do Instead
Look, I’m not here to scare you off AI tools entirely. They’ve genuinely made financial literacy more accessible than at any point in history. For a first-generation university student trying to understand what an index fund is, or someone in a country where financial advisors are expensive and inaccessible — AI is a real leveller.
But there are moments when you need a human with legal accountability. Buying property. Setting up inheritance planning. Dealing with significant debt restructuring. Starting a business. In those cases, the confidentiality obligations a licensed financial planner or advisor carries — obligations that are legally enforceable — matter. An AI chatbot has no such obligations. That’s not a criticism of the technology. It’s just a fact you should factor in.
The short version: use AI to get smarter about money. Use humans to make the biggest money decisions of your life. And in between — check that privacy setting. Right now, if you can.
Last updated: June 03, 2026