OpenAI has introduced a suite of personal finance tools within ChatGPT, initially available to U.S.-based Pro subscribers. The feature, launched on Friday, allows users to securely link their bank and investment accounts through Plaid, granting the chatbot read-only access to balances, transactions, investment holdings, and liabilities such as mortgages or credit card debt.
How it works: Users can connect accounts from over 12,000 supported financial institutions, including major names like Charles Schwab, Fidelity, JPMorgan Chase, Robinhood, American Express, and Capital One. After linking, ChatGPT presents a dashboard displaying portfolio performance, spending patterns, active subscriptions, and upcoming payments. Users can then ask personalized questions such as identifying forgotten subscriptions, comparing monthly expenses, estimating credit card payoff timelines, or planning savings goals.
The integration is powered by Plaid, which handles authentication and data connection between ChatGPT and the financial accounts. OpenAI emphasized that the chatbot cannot initiate transfers, see full account numbers, or make any changes. Users retain control, with the ability to disconnect accounts at any time and have their synced data deleted within 30 days. A toggle also allows users to opt out of having their financial conversations used for model training.
Underlying technology: The feature relies on OpenAI’s GPT-5.5 model, which the company says offers stronger contextual reasoning for finance-related questions. OpenAI developed a new benchmark in collaboration with finance professionals to ensure high-quality responses. Future updates include integration with Intuit, enabling ChatGPT to estimate tax implications of stock sales.
Priced at $200 per month, the personal finance tools are currently exclusive to ChatGPT Pro users on web and iOS. OpenAI plans to gather feedback before expanding access to its Plus tier and eventually all users. The move marks a significant step in blending generative AI with everyday financial management, though privacy and data handling questions linger.
