OpenAI launches ChatGPT finance assistant with access to bank accounts

OpenAI has introduced a new finance assistant inside ChatGPT. Users can connect their bank accounts, credit cards, investment portfolios and loans, after which ChatGPT can answer questions based on a person's real financial data. The feature launched on May 15 as a preview for US Pro subscribers.

How to connect your accounts

The integration runs through Plaid, a platform that has long connected apps to banks. The process is simple: the user opens ChatGPT, selects the option to connect financial accounts, signs in through their bank, chooses which accounts to allow, and that's it. ChatGPT gets access to balances, recent transactions, credit limits and investment portfolios.

For now, the feature is available only to US Pro subscribers on the web and iOS. Expansion to other countries and platforms is planned for later, but there is no timeline.

What ChatGPT can do with your data

Once a bank is connected, new capabilities appear:

• Spending analysis by category and budget recommendations
• Pattern detection: where large sums leak away, which subscriptions pile up
• Planning savings and investment goals
• Quick answers to questions about balances and recent payments
• Analysis of loan terms and debt optimization recommendations
• Tax and investment advice based on the portfolio

In theory, this sounds like a personal financial adviser who knows your numbers. In practice, it is a large-scale system that gathers the most sensitive data of millions of people in one place.

Security and new risks

The idea is appealing: a financial assistant with full access to your accounts. But it significantly raises the risks. If OpenAI or Plaid is breached, criminals would get details of every bank account, loan, investment and payment. This is not just an account password; it is each user's complete financial history.

OpenAI states that data is encrypted end-to-end and that the system follows PCI-DSS standards and other security requirements. But once millions of people connect their accounts, it will become target number one for hackers and state intelligence services. History shows that even the largest companies with huge security budgets suffer leaks.

There is a second problem: data ends up in model embeddings. OpenAI may use anonymized, aggregated financial data to improve its models and recommendations. Users should understand that their financial habits become part of AI training.

What this means

AI is moving from a question-answering helper to a personal financial manager with full access to accounts. That is convenient, but it concentrates risk in one place. Every user should honestly ask themselves: how much do I trust OpenAI with my banking data? And even if you trust the company, you are betting alongside it on flawless infrastructure security and on no breaches.