ChatGPT becomes your banker: OpenAI launches personal finance management for Pro users
🔎 OpenAI no longer just wants to answer your questions — it wants to manage your money
May 2026 marks a turning point for ChatGPT. OpenAI just announced a personal finance experience integrated directly into its assistant, allowing US Pro users to connect their bank accounts via the Plaid network. More than 12,000 financial institutions accessible in a few clicks, with a natural language analysis of your income, expenses, and investments.
This is not a gadget. It is the confirmation of OpenAI's super-app strategy, which is gradually transforming ChatGPT into an all-in-one platform. After health, phone calls, agentic coding with Codex dans ChatGPT Mobile, here comes personal finance. Each new domain conquered extends the time spent within the ecosystem — and locks the user in a little more.
The timing is not coincidental. OpenAI acquired Hiro Finance in April 2026, an AI startup specializing in personal financial advice, backed by Ribbit Capital and General Catalyst. The team of around ten people has been integrated into the product. Months later, the result is here. And for fintechs that thought their technological moat would protect them, the signal is alarming.
The essentials
- ChatGPT is integrating a personal finance experience for US Pro subscribers only, via a partnership with Plaid.
- GPT-5.5 Thinking, the underlying model, outperforms previous models on complex financial tasks according to OpenAI.
- OpenAI acquired Hiro Finance in April 2026 to accelerate the development of this feature.
- Over 12,000 connectable financial institutions, with a guided flow by Plaid for account linking.
- No availability date in France or Europe for now.
- Financial data is processed with specific safeguards detailed in OpenAI's official documentation.
Recommended tools
| Tool | Main use | Price (May 2026, check on openai.com) | Ideal for |
|---|---|---|---|
| ChatGPT Pro | AI personal finance | 200 $/month | US users with advanced financial analysis needs |
| Plaid | Banking connector (integrated) | Free via ChatGPT | Secure connection to bank accounts |
| Claude Opus 4.7 | Alternative financial analysis | 20 $/month (Pro) | Users looking for an alternative without bank connection |
How ChatGPT Finances actually works
Access is available in two ways: type @Finances, connect my accounts in the conversation bar, or click the "Finances" option in the ChatGPT sidebar. A Plaid-guided flow then opens to link your bank accounts, credit cards, and investment accounts.
The process relies entirely on Plaid's infrastructure, the financial connector that already powers the majority of US fintechs (Venmo, Robinhood, Cash App). You select your bank, authenticate with your usual credentials, and Plaid establishes a secure read-only link. ChatGPT never receives your bank passwords.
Once the accounts are connected, GPT-5.5 Thinking takes over. This is the model that makes the difference compared to a simple API integration. OpenAI claims that GPT-5.5 Thinking outperforms its predecessors on complex financial tasks: expense categorization, trend identification, and savings scenario simulations. The model can cross-reference your data with general financial knowledge to produce contextualized insights.
OpenAI's official documentation specifies that financial data is processed under the existing privacy policy of the Pro plan, with additional safeguards. Transactions are analyzed in real time but are not stored indefinitely in the conversation history in an unencrypted manner.
The super-app strategy: every domain conquered is a locked-in user
ChatGPT no longer looks like a chatbot. It looks like WeChat.
The addition of personal finance fits into a clear logic: transforming a question-and-answer tool into a centralized platform. Code, health, calls, now money. Each new vertical reduces the need to leave ChatGPT for a third-party app. This is exactly the model that made Asian super-apps successful.
OpenAI is not hiding this ambition. The acquisition of Hiro Finance, confirmed by TechCrunch in April 2026, shows that the company is not waiting for partners to build these features. It acquires them and integrates them. The Hiro product was shut down on April 20, 2026, its team of ~10 people joined OpenAI, and a month later the result is integrated into ChatGPT.
This strategy comes at a cost: $200/month for the Pro plan. But for OpenAI, the goal is not volume in this segment. It is to validate the model, to prove that users are ready to entrust their most sensitive data to a general-purpose AI assistant. If it works with finance, the next logical step is even more sensitive.
Monetization goes beyond the subscription. The arrival of ChatGPT Ads, la publicité ciblée opened to all US advertisers, raises an inevitable question: an assistant that knows your finances and sells targeted advertising, what exactly is that? OpenAI maintains the separation between financial connection data and the advertising system, but the boundary is thin.
Plaid as a pivot: why OpenAI didn't reinvent the wheel
The choice of Plaid is not a minor detail. It is the most important strategic decision of this launch.
Plaid is the invisible infrastructure of American digital finance. Founded in 2013, bought by Visa for $5.3 billion in 2020 (before being forced to split for antitrust reasons), Plaid connects more than 12,000 financial institutions. When you link a bank account to Venmo, that's Plaid. When you connect your bank to a budgeting app, that's Plaid.
By relying on Plaid, OpenAI avoids the regulatory and technical nightmare of negotiating individually with thousands of banks. The network already exists, the data flows are standardized, the security is proven. Plaid builds the bridge, ChatGPT does the analysis.
On the Plaid blog, the company presents this partnership as the natural convergence of AI and trusted financial data. The argument is solid: data without intelligence = spreadsheets. Intelligence without data = generic advice. The two together = truly personalized experiences.
But this choice also implies a dependency. If Plaid raises its prices, if Visa restructures again, if a regulator questions the data aggregation model, ChatGPT Finances is directly impacted. OpenAI does not control the data access layer, and in finance, whoever controls access controls the power.
GPT-5.5 Thinking: the model that makes financial analysis credible
Financial integration doesn't rely on just any model. OpenAI chose GPT-5.5 Thinking, which ranks second in the agentic leaderboard with a score of 98.2, just behind its own standard version. It is OpenAI's most capable model for complex reasoning tasks.
In the overall leaderboard, GPT-5.5 scores 91, behind Google's Gemini 3.1 Pro (92) but tied with GPT-5.4 Pro. The difference isn't in the raw score but in chain-of-thought reasoning capability, which is essential for financial analysis. Understanding that a series of small-amount transactions over several days corresponds to a recurring subscription, not one-off expenses — that requires contextual reasoning.
OpenAI claims that GPT-5.5 Thinking specifically outperforms previous models on complex financial tasks. The company does not publish a detailed benchmark for this claim, which is notable. But the choice to dedicate a "Thinking" model (with extended chain of thought) rather than a standard model suggests that the complexity of financial data indeed requires deeper processing.
For the user, the result manifests in the quality of the insights. Not just "you spent $340 on restaurants this month" but "your restaurant spending increased by 23% compared to your 6-month average, primarily on Friday nights — here are two scenarios to get back to your budget goal".
Availability: a US-only launch that speaks volumes
ChatGPT Finances is available exclusively for Pro subscribers in the United States. Period.
No date for Europe. No date for France. MacGeneration confirms it: OpenAI is connecting ChatGPT to your accounts via Plaid, but only for US Pro users. The reason is not technical — Plaid works in other countries. The reason is regulatory.
The United States has a relatively permissive framework for financial data aggregation, supported by the Consumer Financial Protection Bureau (CFPB) data access rule. Europe, with the GDPR and the PSD2 directive, imposes much stricter constraints on consent, data retention periods, and the purpose of financial data processing.
OpenAI is playing it safe. Mishandling financial data in Europe would result in fines proportional to global revenue, not just the financial segment. And OpenAI's reputation regarding privacy is not unblemished — US Senate hearings and FTC investigations have left their mark.
For French-speaking users, the message is clear: do not expect this feature before 2027 at the earliest, and expect it to potentially operate differently from the American version. If you want to explore the analytical capabilities of ChatGPT free, Gemini or other free LLMs, it is possible — but without direct banking connectivity.
Privacy risks: your financial data in an AI model
This is the central point of tension. You are entrusting an AI with your income, your expenses, your purchasing habits, your investments. Even in read-only mode, even with encryption, even with Plaid as an intermediary — the amount of sensitive information is considerable.
Blog Nouvelles Technologies analyzes the privacy implications in detail. The fundamental problem is not so much data theft (Plaid has a good security track record) as the use made of it within the broader OpenAI ecosystem.
OpenAI specifies in its help documentation that financial data is used to provide the service and is not used for model training. This is an important commitment, but one that relies on trust in OpenAI's internal processes. There is no public independent audit of this separation.
Three concrete risks:
The risk of cross-profiling. ChatGPT already knows your professional queries, your code projects, perhaps your health data. Adding financial data creates a user profile of unprecedented richness. Even without targeted advertising on financial data, the overall profile is enriched.
The risk of contextual leakage. Language models can sometimes reveal information from one context to another, even with safeguards. A poorly formulated prompt could theoretically lead the model to reference data from a financial session in an unrelated conversation.
The risk of dependency. Once your accounts are connected and your analysis habits are established in ChatGPT, the switching cost becomes enormous. This is exactly what OpenAI wants.
Fintech disruption: when your competitor is a language model
Personal finance fintechs — Mint (closed), YNAB, Copilot, Monarch Money — built their value on two things: data aggregation and analysis. ChatGPT Finances tackles both simultaneously.
Aggregation is delegated to Plaid, which levels the technical playing field. Any fintech can use Plaid. Differentiation therefore relied on the interface and analysis. But when analysis is handled by GPT-5.5 Thinking, which can generate natural language insights well beyond what hand-coded rules can produce, the competitive advantage of traditional fintechs crumbles.
PhonAndroid headlines without hesitation: "No more need for an accountant." It's exaggerated, but the direction is correct. For personal budget analysis, expense categorization, savings recommendations — the tasks that make up 80% of personal finance app usage — a capable LLM does the job.
Fintechs aren't dead yet. Their strength lies in specialization: regulatory, tax, investment. YNAB excels in budgeting methodology. Copilot in the native mobile interface. These fintechs could themselves integrate LLMs to counterattack. But they lose the advantage of AI as a differentiator — since the same model is available in ChatGPT.
The real danger is strategic. Fintechs attract users that ChatGPT can now retain within its ecosystem. Every user who manages their budget in ChatGPT is a user who won't install a personal finance app. It's the classic network effect of a super-app.
What this means for the future of ChatGPT as a platform
ChatGPT Finances is not a feature. It's a prototype of what ChatGPT will be in three years.
The pattern is now clear: OpenAI identifies a vertical domain (health, code, finance), acquires or develops specific expertise (Hiro Finance for finance, expert AI avatar creation capabilities for personalization), integrates via an infrastructure partner (Plaid), and deploys on the US Pro plan before expanding.
The question is: what is the ceiling of this strategy? Personal finance is the most sensitive vertical after health. If users are willing to connect their bank accounts to ChatGPT, there is practically no psychological barrier left for other domains. Real estate? Insurance? Taxes?
The other question is the business model. At $200/month, the Pro plan is a premium product. But the super-app ambition requires a massive user base. Will OpenAI introduce an intermediate tier? Will it monetize financial insights (product recommendations, redirection to partners)? The launch of ChatGPT Ads suggests that advertising will be part of the equation eventually.
For the AI industry more broadly, the signal is that the battlefield is no longer model quality — Claude vs ChatGPT, ChatGPT vs Gemini, the scores are tight. The battlefield is the ecosystem. The one who integrates the most vertical features into its interface wins. Google knows this well — it's exactly what made the success of Android and Google Search.
❌ Common mistakes
Mistake 1: Confusing financial analysis and regulated financial advice
ChatGPT Finances analyzes your data and produces insights. It does not give you certified investment advice. OpenAI is clear about this in its documentation. If you make a major investment decision based on a ChatGPT summary, it is at your own risk. The line between "analysis of your expenses" and "financial advice" is thin, and US regulators (SEC, FINRA) are watching it closely.
Mistake 2: Thinking that financial data feeds model training
OpenAI states that financial connection data is not used for training. Many users assume that everything they type into ChatGPT eventually feeds a future model. This is false in this specific case, but the confusion is understandable given OpenAI's track record on the subject. Always check the current policy, not the one you think you know.
Mistake 3: Underestimating the attack surface risk
Every connected account is an additional attack surface. Plaid is secure, OpenAI is secure, but the entire chain (your device → ChatGPT → Plaid → your bank) multiplies the potential points of failure. Enable two-factor authentication everywhere, monitor authorized access in your bank's settings, and revoke the connection if you no longer use the service.
❓ Frequently Asked Questions
Is ChatGPT Finances available in France?
No. As of May 2026, the feature is reserved for Pro subscribers in the United States. No launch date in Europe has been announced, likely due to GDPR constraints and the PSD2 directive on access to financial data.
Can ChatGPT perform banking transactions?
No. The connection via Plaid is read-only. ChatGPT can analyze your transactions, categorize your spending, and generate reports, but it cannot move money, pay bills, or execute stock orders.
Are my financial data used to train OpenAI models?
According to OpenAI's official documentation, data from the Plaid financial connection is not used to train models. It is processed solely for the purpose of providing the personal finance service.
Which AI model is used for financial analysis?
OpenAI uses GPT-5.5 Thinking, its top-performing model in agentic reasoning (score 98.2), specifically chosen for its ability to handle complex financial tasks requiring in-depth chain-of-thought reasoning.
How much does access to ChatGPT Finances cost?
Access is included in the ChatGPT Pro plan at $200/month (May 2026, check on openai.com). There is no specific additional cost for the financial feature, but it is not available on the Free or Plus plans.
Can I use another LLM to do the same thing?
Not with the same direct integration. Anthropic's Claude Opus 4.7 or Google's Gemini 3.1 Pro excel at financial data analysis if you feed them manually (CSV exports, statement copies). But none currently offer an automated banking connection via Plaid equivalent to ChatGPT's.
✅ Conclusion
ChatGPT Finances confirms what OpenAI's trajectory suggested: the general-purpose AI assistant is becoming a vertical platform that eats fintechs from the inside out. With Plaid as infrastructure and GPT-5.5 Thinking as its analysis engine, the product is technically solid. But the real question isn't technical — it's whether users are ready to entrust their most intimate financial data to a company whose business model is shifting toward targeted advertising. For now, only US Pro users can answer. The rest of the world watches.