OpenAI has spent the past several months signaling that ChatGPT's ambitions extend well beyond conversation. The Operator product executes tasks across the web. Codex writes and deploys code. The memory feature accumulates context across sessions. And on May 15, 2026, the company added one of the most commercially significant capabilities yet: the ability for ChatGPT to read your bank account.

OpenAI launched ChatGPT for personal finance, allowing US users to securely connect their bank accounts through a third‑party financial data aggregator and receive AI‑powered analysis of their spending patterns, savings progress, and financial goals. The launch has been actively covered in today's technology press and represents a direct commercial expression of the Hiro Finance acquisition that OpenAI confirmed in April.

Hiro Finance was built by Ethan Bloch, a serial entrepreneur who previously sold Digit, the automatic savings app, to Oportun for approximately $230 million in 2021. Bloch built Hiro specifically as an AI personal CFO for individuals: users connected their financial accounts, and Hiro's AI modeled different financial scenarios to help them make decisions about savings, debt, and major purchases. The acquisition brought Bloch and a team of approximately ten people into OpenAI, along with the product architecture and financial data infrastructure Hiro had built.

The ChatGPT personal finance feature operates through a secure connection facilitated by Plaid or a comparable financial data aggregator, technologies that have been powering fintech account connectivity for years and whose security standards are well‑established. Users authorize the connection, and ChatGPT receives read‑only access to transaction history and account balances. No money can be moved through the integration. The AI analyzes the data and provides conversational responses to questions that individuals have historically needed either expensive financial advisors or time‑consuming manual analysis to answer.

The kinds of questions the feature addresses include: How much did I spend on food in March compared to February? I am trying to save $20,000 for a house deposit in two years ‑ is my current savings rate on track? Which of my subscriptions am I not using regularly? If I reduced my dining‑out spending by 30 percent, how would that change my timeline to paying off this credit card? These are not simple lookup questions. They require combining transaction data with financial modeling in ways that current personal finance apps do only partially and that most users do not do at all because the process is too time‑consuming.

The privacy implications are significant and have already generated commentary from cybersecurity researchers and consumer advocates. OpenAI's privacy policy for the feature specifies that financial data connected through the bank integration is used to provide the requested financial analysis and is not used to train future models without explicit user consent. The company has not published the full technical specification of how the data is handled, stored, and secured, and the coming weeks will likely produce both detailed privacy assessments from security researchers and regulatory questions from the Consumer Financial Protection Bureau and state regulators who oversee financial data handling.

OpenAI does not hold a financial advisor license, which means the advice ChatGPT provides is legally categorized as general financial information rather than personalized financial advice. The practical distinction between "here is what your spending data shows" and "here is what I recommend you do about it" is one that the feature will need to navigate carefully, particularly for users making significant financial decisions based on ChatGPT's analysis.

The personal finance feature is available to US users on Plus and Pro plans on rollout, with broader availability expected over the coming weeks. Whether it makes OpenAI's platform meaningfully stickier for the segment of its user base that is not primarily using ChatGPT for work or coding, and whether it generates the kind of daily engagement that personal finance apps require to become genuinely useful, are the commercial questions the next few months will answer.