Artificial Intelligence Enters Bank Accounts: How It Works

Artificial intelligence is officially entering the bank account. Last week, the digital bank One Zero announced a new option for its customers: connecting the bank's financial AI agent ("Ella") directly to OpenAI models, allowing customers to consult with ChatGPT regarding their current account data without the need for manual entry or file uploads.

One Zero is not the only financial entity to implement this technology. Over the past month and a half, the investment house IBI, through its IBI SMART app, and the fund management company Fair have also launched similar AI integrations for their customers' investment portfolios. These moves reflect a broader global trend, including companies like Robinhood, driven by the realization that financial institutions must provide a controlled and secure channel for users who are already utilizing AI as an advisor.

Benefits and Functionality

The common denominator for these entities is the desire to translate complex financial data into natural language. In the banking sector, the focus is on analyzing current account behavior and household budgets. Customers can ask questions about spending trends, mortgage repayment changes, or income-versus-expense analysis.

For IBI and Fair, the emphasis is on securities portfolios and financial assets. The integration allows for the analysis of risk distribution, sector exposure, and the impact of market events on specific holdings. "The goal of integrating AI is not to create a gimmick, but to meet the investor at eye level—to lower the fear of financial data and allow them to understand what is really happening with their money in seconds," says Elad Shefer, CEO of Fair.

Technological Approaches

There are two main technological approaches currently operating in the market:

1. The use of the MCP (Model Context Protocol) developed by Anthropic, implemented by One Zero and IBI. This is a secure, unified standard that allows an external AI agent to communicate directly with the bank's systems via an internal financial agent.

2. A dedicated application in the ChatGPT store, as implemented by Fair. Users connect to the official service via a secure identification mechanism.

Security Concerns and Risks

Cybersecurity experts, such as Moshe Karko, CTO of NTT Israel, have warned about the risks of transferring data to American technology companies. Financial institutions counter this with two principles: complete anonymization (no personal identifiers are sent to the AI) and a "Read-Only" mode, which prevents the execution of financial transactions via the chat interface.

Nevertheless, experts emphasize that AI is not an investment advisor. These tools are intended solely for information accessibility, not to replace professional advice. While the industry expects the development of "supervised autonomy" mechanisms in the future, final execution of financial orders will still require confirmation through official banking applications.