After conquering desks and university classrooms, ChatGPT is trying to enter a much more sensitive territory: that of current accounts. A few days ago, Pro subscribers in the United States can connect their bank accounts, credit cards, and investment portfolios directly to the chatbot, obtaining spending analysis, financial planning, and personalized answers about their money.
A step that transforms the conversational assistant into something much closer to a digital financial advisor, with all the opportunities and concerns that follow.
How the integration works
The functionality relies on Plaid, the de facto standard for banking integrations in the United States, which allows connection to over twelve thousand financial institutions, from Chase to Fidelity, including Robinhood and American Express.
Once access is authorized, the user finds a dashboard featuring:
- Portfolio trends;
- Recurring expenses;
- Active subscriptions;
- Upcoming payments.
However, the interaction remains conversational: it is possible to ask ChatGPT why you spent more than usual last month, or to have a five-year plan built to get ready for buying a house.
Technology and Development
Behind the product is also the acquisition of the Hiro team, a personal finance startup acquired by OpenAI last April. According to the company, the new GPT-5.5 model has been calibrated specifically for reasoning in complex contexts such as financial ones, with benchmarks developed alongside industry experts. Integration with Intuit is expected soon, which will allow for tax simulations and preventive creditworthiness evaluations.
Adoption and Privacy
OpenAI states that over two hundred million users already ask financial questions to the chatbot every month. Regarding data management:
- Connections can be revoked from the settings;
- Data is deleted within thirty days;
- Financial memories are manageable by the user.
For now, the service is limited to Pro subscribers on web and iOS, pending extension to Plus users.
The future of chatbots: towards verticalization
Generalist chatbots are fragmenting into specialized verticals, from health to finance—sectors where generic answers are not enough. Anthropic and Perplexity are moving in the same direction, a sign that the next battle will no longer be over the most brilliant model, but over the deepest integration with people's real data.
And here lies the real question that every user would do well to ask themselves before clicking "connect": how willing are we to trust an artificial intelligence that knows every single one of our transactions? The convenience is enormous, but once the perimeter of financial privacy is breached, it is difficult to rebuild.