The mandate. Why the US conversational- finance surface does not translate to Europe.

OpenAI’s new U.S. personal-finance experience in ChatGPT, launched May 15 for Pro users with Plaid account links, is unlikely to carry over to the EU in the same form because European rules treat account-data access, wider financial data sharing and some AI uses in finance as regulated activities, according to a regulatory analysis by Thorsten Meyer AI.

OpenAI said the U.S. preview lets Pro users connect financial accounts in ChatGPT on web and iOS, with support for more than 12,000 institutions through Plaid. The company says users can view spending, subscriptions, upcoming payments, portfolio performance and other account-based information, while asking ChatGPT questions grounded in that data. OpenAI also says the service is not a substitute for professional financial advice.

The confirmed EU regulatory position is different. PSD2 already made account information services and payment initiation services regulated activities. The European Parliament and Council reached a provisional political agreement on PSD3 and the Payment Services Regulation on November 27, 2025; the deal still requires formal adoption before entry into force. Those measures are expected to strengthen the rulebook for payment service providers and open-banking access.

The wider open-finance layer is still pending. The European Commission’s proposed Financial Data Access regulation, known as FIDA, would extend structured financial-data sharing beyond payment accounts to other financial products and create a regulated financial information service provider category. Separately, the EU AI Act lists AI systems used to evaluate creditworthiness or establish a credit score as high-risk, meaning a finance assistant that moves from budgeting support toward credit evaluation could face stricter duties.

Why It Matters

The gap matters for banks, fintech firms, AI companies and users because the U.S. model is built on private data-aggregation rails, while the EU model is built around regulated access, permission management and supervisory oversight. In the United States, the user-facing product can sit on top of an existing aggregator connection. In the EU, the access rights, consent flows, data scope, API quality and AI classification may shape the product before the chat interface is designed.

That changes who is best placed to compete. Firms with financial licenses, regulated data-access operations, consent infrastructure and relationships with supervisors may have an advantage over companies that won in the U.S. by moving quickly on private aggregator coverage. The issue is not only whether OpenAI can localize a feature; it is whether any AI company can offer a full-picture finance assistant in Europe without building around financial regulation from the start.

Background

Open banking in Europe began as a legal mandate rather than a purely commercial market. PSD2 required banks to provide access to account data for licensed third-party providers when customers gave permission. The coming PSD3 and PSR package is meant to update those rules and make payment regulation more uniform across the bloc.

FIDA would take the same logic into open finance, covering data categories such as investments, pensions, insurance, mortgages and loans if the final legislation keeps the broad approach proposed by the Commission. Thorsten Meyer AI argues that this means a European conversational-finance product would need to be built around licensing, consent dashboards, API conformance and AI-risk classification rather than treating those as later compliance tasks.

“We’re releasing a preview of a new personal finance experience in ChatGPT to Pro users in the U.S.”

— OpenAI

“ChatGPT can help you stay informed, but it is not a replacement for professional financial advice.”

— OpenAI

“Parliament and the Council reached a provisional political agreement on PSR and PSD3.”

— European Parliament legislative material

“Europe has no permissionless substrate. It has a mandate at every layer.”

— Thorsten Meyer AI

What Remains Unclear

It is not yet clear whether OpenAI plans to launch the finance experience in Europe, what licenses or partnerships it would use, or whether the product would be narrowed to avoid regulated credit or advice functions. FIDA’s final text and application dates also remain unsettled. The AI Act classification would depend on the product’s intended use and how its outputs are used by consumers, lenders or other financial firms.

What’s Next

The next milestones are formal adoption and publication of PSD3 and PSR texts, continued FIDA negotiations, and AI Act implementation for high-risk systems and finance-linked use cases. Any company planning an EU conversational-finance product will need to map data access, user consent, licensing status and AI risk before launch.

Key Questions

What did OpenAI launch in the United States?

OpenAI launched a personal-finance preview in ChatGPT for U.S. Pro users on May 15, 2026. It lets users connect accounts through Plaid and ask questions based on their financial data.

Why would the EU version be harder to launch?

EU rules treat access to bank-account data as a regulated activity, and proposed open-finance rules would extend that model to more financial data. A provider may need licenses, formal consent controls and supervisory compliance before offering a similar service.

Does the EU AI Act block personal-finance chatbots?

No. But AI systems used to evaluate creditworthiness or establish a credit score are listed as high-risk. A budgeting chatbot is different from a tool that influences credit decisions, and the boundary would depend on product design and actual use.

Is FIDA already in force?

No. FIDA is a proposed EU regulation for financial-data access. Its final form, fees, scope and application dates remain subject to the legislative process.

What is the main takeaway for users?

A U.S.-style account-linking finance assistant may appear simple to users, but in Europe the same idea sits inside banking, data-access and AI regulation. That could affect availability, features, consent screens and who is allowed to operate the service.

Source: Thorsten Meyer AI