OpenAI launches ChatGPT for financial services

OpenAI is introducing ChatGPT for Financial Services, a version of ChatGPT Work designed for banks and other financial institutions. The service combines integrated financial data with the reasoning capabilities of GPT-6 Astra to create research, financial models and client materials.

The product grew out of a design collaboration with Morgan Stanley and Evercore. That work helped OpenAI identify two common problems for investment banking and equity research teams: finding reliable data without wasting time and turning analysis into usable documents, spreadsheets and presentations.

Financial data ready to use

ChatGPT for Financial Services includes data from providers such as Daloopa, PitchBook, LSEG News and Crunchbase. The content covers, among other things:

• Earnings call transcripts
• Financial statements and fundamental data
• Information on private companies
• News and market data

OpenAI indexes and hosts this data on its own infrastructure. The company says this improves response speed, information retrieval and how sources are displayed.

One important feature is granular citations. Analysts can trace a figure or claim back to the source document and review the evidence while preparing their analysis.

For example, a banker normalizing an income statement can review which costs were excluded to calculate adjusted EBITDA, consult the reconciliation notes and decide how to use that figure in a valuation. The tool does not eliminate professional review, but it reduces some of the manual work required to get there.

Analysis and documents in the same workspace

GPT-6 Astra is natively integrated and geared toward three common tasks in finance:

• Searching for and retrieving information from complex documents
• Interpreting figures, tables, notes and financial periods
• Turning analysis into documents, spreadsheets and presentations

The tool can compare data from different sources and periods, interpret annotations in public documents and generate interactive charts. Teams can then turn that work into valuation models, research reports or client presentations.

Administrators can also publish Excel, Word and PowerPoint templates. This allows each team to work with its firm's format, style and brand guidelines instead of rebuilding every document from scratch.

Connections to the systems banks already use

Institutions that already subscribe to other providers will not necessarily have to purchase the data again. OpenAI is working with S&P Capital IQ, LSEG, MSCI, Dow Jones Factiva and Moody’s to enable shared login systems and permission recognition.

The idea is that a provider can identify the user through their ChatGPT account and automatically give them access to the data they are already authorized to use. The service also includes more than 50 connectors, including Datasite, Box, Preqin, FactSet and Intapp.

These connectors use MCP, a standard that connects AI models with external tools and sources. OpenAI says it has optimized some of the most widely used connectors in finance so teams spend less time troubleshooting configuration issues and more time reviewing results.

Controls for confidential information

In finance, security is not an add-on. Banks work with customer information and, in some cases, material nonpublic information, meaning data that could affect the market but has not yet been disclosed.

ChatGPT for Financial Services builds on ChatGPT Enterprise controls, including:

• Enterprise login with SAML SSO
• User provisioning through SCIM
• Role-based permissions
• Encryption of data in transit and at rest
• Administrator-configured retention settings
• Log exports for audits and investigations

Each firm's business data is not used to train the models by default. Companies can also control which applications and read or write actions are available to each role, and create multiple workspaces to keep information separate between teams.

The service is available to financial institutions that meet OpenAI's requirements. For now, the initial focus is investment banking and equity research, although the company plans to expand coverage to other areas of the industry.

For you, the main change is concrete: tasks that currently require searching multiple databases, checking figures and adapting documents can be brought together in one environment. The question to watch is not only whether AI writes faster, but whether analysts can verify every piece of data and maintain control over sensitive information.