Banks scramble to respond to autonomous AI hacking threats

The OpenAI logo is seen in this illustration created June 11. Reuters-Yonhap

Banks and other financial companies are strengthening their cybersecurity by deploying artificial intelligence (AI)-powered security tools as the emergence of autonomous AI agents raises concerns over sophisticated cyberthreats, industry officials said Wednesday.

Officials say that the financial sector is entering an "AI versus AI" era, where AI is increasingly being used not only to launch cyberattacks but also to detect vulnerabilities in advance and strengthen defenses.

Major financial groups and their affiliated banks have recently accelerated the adoption of AI-based cybersecurity tools.

Woori Financial Group launched its AI-powered penetration testing platform Xint in May.

Xint uses AI to identify complex security vulnerabilities by replicating the techniques of ethical hackers. The platform has reduced security assessments that previously took about two weeks to less than 12 hours, enabling continuous testing, the company said.

"Rather than simply scanning for known vulnerabilities, the platform uses AI trained on the attack methods and analytical techniques of ethical hackers to identify complex attack paths that would be difficult for humans to detect," a Woori official said, adding that the platform has already detected vulnerabilities that had not been identified by previous security tests.

Hana Financial Group said it is developing its own Hana AI Security Framework (HASF), which applies AI throughout the cybersecurity process, with a phased rollout underway.

The group also became the first financial institution selected to participate in the government's regulatory tests allowing the use of generative AI for cybersecurity purposes.

Shinhan Bank, for its part, has been using generative AI since June to conduct penetration tests and strengthen its threat detection and response capabilities across its systems. The bank said the program is aimed at strengthening its vulnerability detection and incident response capabilities using AI.

KB Kookmin Bank said it is building an AI-based security inspection system to strengthen customer data protection and enhance cybersecurity, with an aim to develop more advanced technologies to respond to AI-powered cyberthreats.

A row of ATMs belonging to major commercial banks is seen in Seoul, July 22. Yonhap

The series of moves comes as leading AI developers have begun demonstrating autonomous AI agents capable of independently identifying vulnerabilities and carrying out hacking-related tasks without human intervention.

OpenAI recently disclosed that its advanced AI models escaped a sandbox testing environment, accessed the open internet and autonomously hacked into the infrastructure of AI platform Hugging Face.

Earlier this year, Anthropic also raised concerns across the cybersecurity industry after demonstrating an autonomous AI system capable of bypassing security defenses with far less human intervention than previous AI models.

While no major incidents involving autonomous AI hacking have been reported in Korea's financial sector so far, industry officials said early preparation is essential because the country's highly digitalized financial system makes it an attractive target for cyber criminals.

"The widespread use of digital banking and integrated mobile financial services has significantly increased the amount of sensitive personal data handled by financial institutions, making cybersecurity increasingly critical," a banking industry official said.