ChatGPT enters the heart of finance. OpenAI introduced ChatGPT for Financial Services, a version of its platform designed for investment banks, finance firms and equity analysts.
This is not an application that can independently access the current account or invest savers’ money. It is a professional tool with which operators can search for financial information, compare financial statements, build models and prepare documents for clients.
Artificial intelligence could, in fact, make some banking services faster and reduce the time needed to analyze companies and investments, but it does not automatically guarantee better advice, lower commissions or higher returns.
What is ChatGPT for Financial Services
According to OpenAI, ChatGPT for Financial Services was developed with Morgan Stanley and Evercore as design partners and is powered by GPT-6 Astra.
The system integrates information from professional financial providers such as LSEG, PitchBook, Daloopa, Crunchbase and Quartr. You can, therefore, consult company data, balance sheets, transcripts of results conferences and company information.
Companies that already have professional subscriptions can also connect to services like FactSet, S&P Global, Preqin and Datasite. It’s a crucial detail: the value of the platform comes not just from the model’s ability to write, but from organized access to verifiable and up-to-date financial data.
The leap compared to the ChatGPT normally used by the public is, therefore, above all, in data integration and company controls. The platform is built on ChatGPT Enterprise security measures, including role-based access, encryption, and activity log export for compliance checks.
What changes for those who invest
For a saver the first effect could be greater speed. An advisor could obtain in a few minutes comparisons between funds, stocks, bonds and economic scenarios that today require several hours of work.
AI could also be used to produce more personalized reports, quickly update a portfolio’s profile and explain concepts such as volatility, diversification and risk in simpler language.
This does not mean, however, that ChatGPT will be able to replace the evaluation required for investment advice. ESMA has made it clear that financial companies remain responsible when using artificial intelligence systems and must continue to act in the best interests of the customer.
In other words, if a suggestion produced by the algorithm turns out to be wrong, a bank cannot defend itself by claiming that “the artificial intelligence decided it”.
The possible effects on consultants and analysts
The most exposed jobs are not necessarily those that require final decisions, but those built on manual data collection and sorting.
A junior analyst may spend many hours copying numbers from financial statements, updating spreadsheets, and adapting presentations. ChatGPT for Financial Services targets these businesses directly.
The consequence could be an increase in productivity, but also a transformation of entry paths into the sector. In fact, there is a paradox: the most repetitive activities are also those through which young people learn to read a balance sheet, recognize an error and understand the structure of an operation.
If these tasks are immediately entrusted to artificial intelligence, banks and financial companies will have to rethink training. The risk is not only losing some junior positions, but finding ourselves in a few years with fewer professionals capable of controlling what the machine produces.
Will ChatGPT be able to decide how to invest customers’ money?
The answer, at the moment, is no. The platform can process information and produce analyses, but the final decision and related responsibility remain with the financial company and the professionals who use the system.
It should also be distinguished from the personal finance function available in the consumer-facing ChatGPT. ChatGPT for Financial Services is a separate plan intended for financial institutions: it does not imply that OpenAI automatically knows the balance, movements or investments of a bank’s customers.
Any access to the data will depend on the integrations authorized by the institute, the permissions assigned to the operators and the rules applicable to the protection of information.
What are the risks to customer data
The security declared by the platform does not eliminate all risks. A bank will need to establish what information can be used, who can view it and how long it must be kept.
The main points to check are:
- the possible insertion of confidential or unnecessary data
- the possibility that a response contains incomplete information
- the use of outdated or incorrectly interpreted sources
- the traceability of changes made by operators
- dependence on a single technology supplier
- the ability to reconstruct how a recommendation was produced.
The European Union’s AI Act establishes a framework based on the risk level of systems. Furthermore, the regulations on artificial intelligence continue to be accompanied by banking, financial and personal data protection regulations.
Better services, but not necessarily less expensive
Automation can reduce the time it takes to prepare an analysis or presentation. There is, however, no guarantee that these savings will be passed on to customers through lower fees.
Banks could use the increased productivity to serve more customers, expand margins or improve the quality of service. Much will depend on the competition and the ability of customers to compare costs and results.
The same goes for investment performance. Having more data and analyzing it faster does not mean predicting market trends with certainty. If numerous companies use similar models and the same sources, it may even increase the risk of very similar decisions being made at the same time.
The real revolution is not having more information
In professional finance, information is already abundant. The change consists in making quantities of data that today require different tools and skills accessible, comparable and usable in a few seconds.
The faster AI becomes at producing an answer, the more valuable the human ability to ask the right question, recognize a fragile hypothesis, and take responsibility for the choice becomes.
ChatGPT could, therefore, reduce the economic value of manual number collection. However, it does not eliminate the need for judgment. For customers and savers the decisive question will not only be “does the bank use artificial intelligence?”, but “who checks the answer and who responds when it is wrong?”.
