Banks race to ramp up AI defenses as deposits fall after cyberattacks
Customers use an ATM at a KB Kookmin Bank branch in Seoul, Sunday. Yonhap
ARTEX developer's update halt unlikely to prevent further attacks
A wave of recent data breaches at major Korean banks is putting pressure on both customer confidence and the industry’s defenses, with deposits at these lenders declining by more than 20 trillion won ($15 billion) in just 11 days, according to industry officials, Sunday. They are now stepping up security spending and hiring specialists to keep pace with increasingly sophisticated artificial intelligence (AI)-driven attacks.
Demand deposits at the country’s five largest lenders — KB Kookmin, Shinhan, Hana, Woori and NH NongHyup — totaled 675.5 trillion won on Sept. 30, but dropped to 655.4 trillion won as of Sunday. Fixed-term deposits, which had risen for five consecutive months, also fell by more than 2.7 trillion won to slightly more than 1 quadrillion won.
On Oct. 1, Shinhan Bank first reported that about 25,000 customers were affected, with names, phone numbers, annual income and loan limits exposed. The following day, KB Kookmin reported 119 affected customers, while Hana Bank said personal information belonging to 89 customers was exposed.
Lenders are responding with bigger security budgets, more specialized staff and greater use of AI to test their own defenses.
KB Kookmin Bank plans to increase its information security budget to more than 100 billion won next year, up from this year’s 86.1 billion won.
Shinhan Bank said it will set next year’s information security budget at “the highest level in the industry.” It has not disclosed a specific amount.
Woori Bank is considering increasing spending on security systems by more than 20 percent from this year and expanding its specialist workforce by at least 10 percent.
Banks are also using their own AI tools to find weaknesses before attackers do.
Woori Bank uses Xint, an AI-powered security testing tool developed by Korean firm Theori, alongside tools developed in-house. Xint generates attack scenarios and runs penetration tests to identify vulnerabilities.
KB Kookmin is testing AI defense technologies with several security firms, including Stealth Solution. Its in-house ethical hackers are also assessing how well active defense systems can respond to different types of AI-driven attacks.
NH NongHyup is set to introduce its own AI red-teaming system in the second half of next year. The system will probe AI models, data and services for vulnerabilities from an attacker’s perspective.
Hana Bank, meanwhile, is working on systems that can respond to hacking attempts on devices and servers in real time. It is also pursuing AI-based cybersecurity and automated penetration testing platforms.
“Because security personnel cannot manually keep up with attacks that use AI, we are working to make greater use of AI in our own defenses,” an official at a major bank said.
Meanwhile, ARTEX, a Chinese open-source AI cybersecurity tool reportedly used in the attacks, is no longer being developed for public release. Its developer, who goes by “Autumn-27,” said on GitHub that the tool had been “abused by some bad actors.”
“Given the misuse of the tool, the ARTEX project will no longer be updated,” the developer wrote. “No further versions will be released to the public.”
This, however, doesn’t affect copies that have already been downloaded and in circulation, meaning the move is unlikely to prevent further attacks using ARTEX. The possibility of similar attacks involving other AI agents has also become a more immediate concern.