5 Things You Should Never Tell Your AI Chatbot (to Keep Your Money Safe)

AI chatbots like ChatGPT, Gemini, and Claude are convenient tools for drafting emails, brainstorming ideas, or summarizing documents. But their usefulness can lead to a false sense of privacy. People often treat chatbot conversations like a private diary, forgetting that anything typed may be stored, reviewed, or even used to train future models.

In a recent column for The Washington Post, personal tech writer Shira Ovide outlined five categories of financial and personal information you should never share with an AI chatbot. The advice is straightforward but easy to overlook in practice.

What happened (and why it matters)

The column wasn’t triggered by a specific data breach or scam, but by growing awareness that chatbot logs can become a liability. If your account is compromised, an attacker could read your entire conversation history. Even without a breach, companies behind the chatbots retain your inputs for moderation, quality assurance, or model improvement. Some have been known to accidentally expose chat logs through bugs or misconfigurations.

The risk is not hypothetical. There have already been incidents where users inadvertently pasted confidential business strategies or personal financial numbers into public-facing AI tools. Scammers are also becoming more sophisticated, sometimes using chatbots themselves to generate convincing phishing messages based on information victims have shared elsewhere.

What you can do: five concrete rules

1. Never share your full bank account or routing numbers.
Even if you’re asking the chatbot to organize your budget or categorize transactions, use placeholder amounts or dummy account labels. A bank account number combined with your name and address is enough for fraudsters to attempt unauthorized transactions or set up automatic payments.

2. Keep passwords, PINs, and security questions out of the chat.
Some people use chatbots as a password manager of sorts, typing in credentials “just to remember them later.” This is dangerous. Not only are these entries stored, but if your chat history is ever indexed or leaked, your passwords become public. Use a dedicated encrypted password manager instead.

3. Do not provide your Social Security number or tax ID.
This might seem obvious, but some users have experimented with “AI tax preparation” or budgeting tools by feeding in their full SSN to get personalized scenarios. No legitimate financial planning requires you to share your SSN with a chatbot. If a tool claims it does, that’s a red flag.

4. Don’t paste your whole credit card number.
Even for a question like “Can you find a subscription charge on this statement?” it’s safer to redact the card number and just describe the vendor and amount. Many chatbots only need context, not the raw digits, to help.

5. Avoid sharing your precise home address or location.
While asking for restaurant recommendations near you is fine, don’t say “I live at 123 Maple Street, Apartment 4B, Springfield.” Your full address is sensitive personally identifiable information (PII). If a chatbot’s logs are later accessed, that address could be used for identity theft, stalking, or targeted scams.

The broader principle

Treat every chatbot interaction as if it could be read by a stranger tomorrow. That doesn’t mean you can’t use them for productivity—just be deliberate about what you type. If a task requires sensitive financial data, do it outside the chatbot window, in tools designed for privacy.

The Washington Post column is a good reminder that convenience shouldn’t come at the cost of security. AI chatbots are here to stay, but your personal and financial data doesn’t have to be part of the training set.

Sources

  • Ovide, Shira. “Don’t tell your AI chatbot these 5 things to keep your money safe.” The Washington Post, April 25, 2026.
  • Additional background on chatbot data storage practices from public disclosures by OpenAI, Google, and Anthropic.