Home > Business > Finance

As AI agents reshape finance, success hinges on balancing innovation and security

Published: 15 Oct. 2025, 07:00

JIN MIN-JI
[email protected]

An AI assistant serves a customer at Shinhan Bank’s AI-powered branch in central Seoul. [SHINHAN BANK]

Picture a future where you no longer need to dig through banking apps to find the lowest rates or wait on hold with customer service. Instead, you simply say, “Switch my credit loan to the lowest interest rate available.” You may also easily manage your spending just by saying, “Cap my credit card spending at $1,000 a month.”

In a matter of seconds, an AI agent-powered platform scans the market, evaluates options and adjusts your financial setup for maximum benefit — all on its own. That is a future that may soon become possible in the era of AI agents: technology capable of independently carrying out tasks and addressing problems.

AI agents are stepping up in the traditionally cautious world of finance, acting as digital co-workers and taking on more sophisticated tasks, from personalized customer service to fraud prevention.

Though concerns about breaches in security persist, experts agree that AI agents will become more widely used — even in the highly conservative world of finance — as their ability to deliver personalized and real-time decision-making increasingly outweighs the perceived risks.

The market size of global AI agents in financial services was valued at $490.2 million in 2024, according to San Francisco-based Grand View Research. The market size is projected to grow from $691.3 million in 2025 to $4.49 billion by 2030 at a compound annual growth rate of 45.4 percent during the forecast period.

“The effectiveness of adopting AI agents in the finance sector is particularly high compared to other industries as most of the data handled by financial companies consists of numerical information, which can be easily standardized,” said Baik Yeon-ju, a research fellow at the Korea Institute of Finance.

An AI assistant is ready to serve a customer at Shinhan Bank’s AI-powered branch in central Seoul. [SHINHAN BANK]

From assistant to lead
Since opening an AI branch in central Seoul last year, Shinhan Bank — the No. 1 commercial bank in terms of net profit last year — is working on adding more complex services with the help of AI agents.

Currently at AI service desks, tasks that are typically handled at traditional teller windows — like currency exchange, check card issuance and electronic financial services — can be processed. But in the coming months, the bank plans on applying a voice-based LLM (large language model) agent capable of understanding and addressing more complex customer needs to process inquiries, action customer requests and make product recommendations.

Shinhan also expects its wealth management AI agent, planned to be developed by March of next year, to expand its private banker services — once available only to a select wealthy few — to all customers. Based on customers' questions or concerns, the AI agent can design personalized financial strategies in real time by considering factors such as life cycle, income, preferences and market conditions. This would enable fully autonomous financial consultation and product design, as well as offer wealth services to customers across a more diverse range of economic groups.

“The wealth management agent will not just be an answering machine but rather a facilitator that coordinates data and functions, understands context and supports decision-making,” said Han Min-seok, the manager of the Customer Solution Group at Shinhan Bank’s AI Research Center.

At KB Kookmin Bank, AI agents are mostly supporting employees behind the scenes.

For employees serving high-net-worth clients, private banking agents generate key materials needed for consultations, including customer analysis, market and investment research and portfolio proposals.

For employees handling loans, relationship management agents assist with tasks such as analyzing target companies that may draw loans, preparing business proposals and supporting loan execution. The content does not come from generic templates but is based on comprehensive assessments of customer risk preferences, expected returns and market conditions.

"As providing misinformation in finance carries high risks, banks are approaching this rollout cautiously," said Lee Yong-gil, a spokesperson for KB Kookmin Bank. "Over time, the roles of AI agents could be extended to supporting headquarters operations, administrative functions, and even areas such as financial planning, accounting and risk management."

The state-run Industrial Bank of Korea currently uses an AI agent that “automatically collects, categorizes and briefs” government policies on its employees. In the future, it plans to deploy AI-driven monitoring and blocking agents to detect and swiftly block fraudulent activities, including voice phishing and financial crimes involving fraudulent bank accounts. In September, the LG Group’s AI lab launched a financial agent in partnership with the London Stock Exchange Group that can analyze data to generate financial reports and make economic forecasts without human intervention.

From helpful to hands-on
Some overseas financial companies are more active adopters of AI agents.

JPMorgan Chase has adopted AI agents across its services, including wealth and asset management, as well as risk and compliance.

For instance, its agent for defense system analyzes transaction data in milliseconds, evaluating the amount and location, as well as subtle behavioral signals like a user’s typing cadence during an online transaction, to spot anomalies. The company’s wealth management agent incorporates predictive AI models that analyze macroeconomic indicators and trading patterns to anticipate clients' needs before they even reach out.

"Our advisers are finding the right information up to 95 percent faster, which means they spend less time searching and more time engaging in meaningful conversations with clients," said Mike Urciuoli, the chief information officer at J.P. Morgan Asset and Wealth Management, according to a Reuters report in May.

Credit rating agency Moody’s has created 35 agents designed to generate comprehensive financial risk reports and other tasks faster than humans. They are capable of analyzing macroeconomic trends and assessing company-specific risks by leveraging proprietary financial data.

In the future, Moody’s expects AI agents to automate credit analysis by analyzing a company’s financial history, industry trends and economic conditions to produce a comprehensive credit assessment. AI agents will also make it possible to enhance risk management by forecasting potential economic downturns by evaluating indicators such as unemployment rates, interest rates and geopolitical events.

Capital One, a major American bank, built a multiagent conversational AI assistant that can mimic human reasoning to not only provide information but act on customers’ behalf based on their requests.

For instance, if a customer wants to buy a car, the agent — supported by Capital One, a provider of auto financing — is designed to guide them through the entire purchase process, from vehicle comparisons to test-drive reservations and dealer appointments, handling these multiple tasks at once.

Some experts say that Korea needs to speed up its adoption of the agents.

“Korea needs to adopt AI agents more aggressively as a second-mover country, but its development is much slower than that of the United States,” said Yoo Byung-joon, a professor at Seoul National University Business School.

“A primary reason for that is Korea’s Personal Information Protection Act,” Yoo continued, noting the act’s lack of clear regulations and standards for handling personal data beyond its original collection purpose.

The hidden costs
As banking becomes more programmable with AI agents, customers who once passively accepted fees and missed out on benefits will be empowered to take control simply by issuing commands to AI agents, negatively impacting financial companies' customer-directed revenue streams.

“The shift to agent-mediated finance is an inflection point. For banks, thriving in this new era means reengineering products around performance, not loyalty,” according to McKinsey & Company report titled "The end of inertia: Agentic AI’s disruption of retail and SME banking."

Agentic AI systems can "monitor balances in real time, compare returns across institutions, sweep idle cash into higher-yield accounts, and then sweep cash back to a checking account in time for bills," it added. "That allows more of the spreads once captured by banks to go to account holders."

At the same time, as these AI systems access highly sensitive financial data, financial institutions are racing to establish robust privacy frameworks to protect customer information.

Woori Bank plans to establish its own guidelines for AI development and utilization and launch a new decision-making body within this year. Shinhan Financial Group already established a group-wide AI governance standard in November last year and is currently working to internalize it across all its affiliates by the end of this year. It plans to establish a security review system for AI services that reflects regulatory AI guidelines, implement filtering to detect whether sensitive or personal information is included in inputs or outputs through AI prompts and detect the presence of personal information within databases used for AI training during the development phase.

“Security is critically important when adopting AI in finance, but there’s an inherent trade-off between strong security and the widespread adoption of AI agents in business,” said Song Min-taek, a professor at Hanyang University’s Graduate School of Business Administration. “In particular, in the financial sector, the country that strikes the right balance between the two will be best positioned to lead the AI agent-driven economy.”

BY JIN MIN-JI [[email protected]]