Industry observers note that so it’s not surprising that OpenAI is zeroing in on personal finance, with ChatGPT now capable of delving into our banking, checking, and investment accounts. If there’s one area where LLMs excel, it’s plowing through reams of data and teasing out patterns, trends, and insights.
In a fresh development, using a secure third-party ecosystem called Plaid (support for Intuit is coming soon), ChatGPT Pro (and eventually Plus) users can enable integrations with more than 12,000 financial institutions, allowing ChatGPT to connect to and sync with a wide range of banking and investment accounts, from Bank of America and Charles Schwab to American Express and Robinhood. (By the way, more than 200 million users a month are already coming to ChatGPT with money questions, OpenAI says.). First, some background.
Industry observers note that chatGPT will be able to generate more personalized responses to queries like “Have I been spending more recently,” “Help me come up with a plan to put away some money each month,” and “What did my recent vacation actually cost?” Those answers can come as standard text replies, tables, bar charts, pie charts—theoretically, you could get a quick snapshot of your financial position in just a few prompts. Once connected, ChatGPT can use the data and transactions pulled from your various financial accounts for a dashboard that summarizes your current financial position, as well as for context in its conversations with you.
Industry observers note that on this point, we can relax. Even after granting access to your accounts, ChatGPT can only look at balances, transactions, investments, and debts; it can’t actually change anything, nor can it see your full account numbers. But the idea of ChatGPT getting access to your bank accounts raises an immediate concern: Could it drain your accounts with one bad prompt or hallucination?
According to the latest update, openAI also says you can disconnect your various banking and investment accounts at any time, and that if you do, your associated financial data will be wiped from OpenAI’s servers within 30 days.
As part of the ongoing story, while you can untether your financial account integrations in just a few clicks, any discussions you’ve had with ChatGPT about your finances will stay in those conversations unless you delete those specific chats manually. ChatGPT may also save “memories” about your finances, which you can manage and/or delete from the Finances page in the Settings menu. Things get dicier in a few other areas.
According to the latest update, you’ll definitely want to tweak that before using ChatGPT’s fresh personal finance integrations—all ChatGPT users should consider making that change. More concerning is the fact that ChatGPT can use its financial discussions with you to train its models, unless you toggle off the “Improve model for everyone” setting in Settings → Data controls.
Industry observers note that the account access is read-only, you can turn off the integrations (albeit with a brief lag time for getting your data wiped from OpenAI’s servers), and financial “memories” are stored in a centralized spot where they can easily be nixed. Aside from the opt-out model training, it does feel like OpenAI has done everything mostly right when it comes to allowing ChatGPT to pick through our books.
The report highlights that there’s also the integrity of Plaid (the financial accounts gateway) to consider, but from all accounts it isn’t any more or less secure than Intuit, which I use to connect my banking and investment accounts to Quicken.
Industry observers note that all those transactions in your banking and investment accounts paint an intimate portrait of who you are, what you do, and where your priorities lie. Yes, you can pick and choose which accounts you allow ChatGPT to see, but ChatGPT’s financial insights won’t be terribly useful unless you give it as much access as possible, and even a single credit card statement can pack a ton of personal details you may not want shared. And yet, I’m not ready to throw open my books to ChatGPT.
According to the latest update, that doesn’t seem likely, and we already know it can’t actually do anything with your accounts beyond examining them. But the way in which ChatGPT can “remember” your financials from previous conversations can be tough to track. (One option would be to confine financial discussions with ChatGPT to temporary chats, which are discarded once they’re closed.). Could ChatGPT somehow run amok and drop my financials to the global stage?
In a fresh development, openAI says that ChatGPT’s financial chats will default to GPT-5.5 Thinking—its most powerful reasoning model—and that it consulted with more than 50 finance professionals to grade ChatGPT’s performance. There’s also the issue of the quality of ChatGPT’s financial advice.
Industry observers note that asking ChatGPT about recipes is one thing, but seeking financial advice is another—a misinterpretation or hallucination can have drastic outcomes. Yet the substance of ChatGPT’s replies can be greatly influenced by the prompts we give it.
Industry observers note that i also realize that when it comes to finances, the privacy ship has pretty much sailed. (What does Bank of America know about me, anyway?) Still, I’m leery of connecting my bank accounts to ChatGPT. Perhaps I’ll be singing a different tune in a year, but I’m just not there yet. Maybe I’m just being paranoid.
As part of the ongoing story, his coverage of artificial intelligence interrogates the most recent LLMs, and how they can be used at work and at home to be best prepared for the AI revolution. “AI is going to change our lives sooner than we think,” Ben writes. “Our best way to adapt is by using it every day.” Ben has been a PCWorld author since 2014, and has covered everything from laptops to security cameras before launching PCWorld’s AI beat. Ben's articles have also appeared in PC Magazine, TIME, Wired, CNET, Men's Fitness, Mobile Magazine, and more. Ben holds a master's degree in English literature. Ben has been writing about consumer technology for more than 20 years, and now focuses his reporting on AI as it relates to the basic human experience.