Illustration – A person is sitting in front of a computer screen. South Korea’s intelligence agency said North Korea’s AI-assisted cyberattacks can cause tens of thousands of malicious acts every second. Sasha Steinbach / EPA
June 10 (Asia Today) — Artificial intelligence is reshaping the cybersecurity battlefield, South Korea’s intelligence agency has warned, saying North Korean hacker groups are moving toward autonomous attacks that can identify vulnerabilities, penetrate systems and monetize stolen data with limited human involvement.
The National Cyber Security Center, which operates under the National Intelligence Agency, issued the warning in its 2026 National Information Security White Paper released on Sunday.
The agency said that while the rapid development of AI is rapidly improving the capabilities of attackers, the spread of cloud infrastructure and neglect of aging systems are exposing structural weaknesses in South Korea’s cyber defenses.
The agency specifically focused on the rise of agent AI, a type of autonomous artificial intelligence that can set goals, analyze data, and manipulate external systems without constant human direction.
This technology can be used by hackers to generate large volumes of phishing messages and other socially engineered content, develop hacking tools such as ransomware, and carry out large-scale operations with fewer people, less time, and at lower cost.
Concerns about the exploitation of agent AI were recently raised after Anthropic’s AI model Mythos was reported to have generated Windows attack code in 31 minutes.
This change is especially noticeable among North Korean hacking organizations. Global cybersecurity companies, including Kaspersky and Google Threat Intelligence Group, have identified signs that the North Korea-linked group Kimsuky was using large-scale language models to help write its code.
APT45, another North Korea-linked hacker group, repeatedly entered extensive prompts to search for software vulnerabilities and test whether their exploit code could be executed.
Analysts increasingly believe that North Korea began designing and testing automated AI attacks last year and has now largely adopted the technology. The changes are expected to allow North Korean hackers to overcome personnel limitations and launch large-scale attacks on a regular basis.
Last year, North Korea stole a record 2.2 trillion won (approximately $1.46 billion) in virtual assets.
While North Korea’s cyber capabilities are rapidly improving, many South Korean public and private systems remain vulnerable due to aging infrastructure. Risks are increasing as organizations deploy AI into more areas of their operations without completely updating their defenses.
The agency said agent AI is particularly suited to manipulating AI systems used by targeted organizations, meaning it could expand potential attack routes unless South Korea strengthens its security systems.
“Starting this year, agent AI will autonomously execute the entire attack lifecycle and generate tens of thousands of malicious actions per second,” the agency said. “Defense systems must also immediately transition to autonomous security operations that identify and isolate threats at machine speed, with minimal human intervention.”
Experts said isolated responses are no longer sufficient and called for the establishment of a national control tower capable of continuous cyber response.
“The only way now is to use AI to discover security issues, patch them as soon as possible, and prevent attacks,” said Choi Byung-ho, a professor at Korea University’s Human-Directed AI Research Institute. “We need a governance system that can respond to hacking within 24 hours, but this is difficult due to issues such as delegation of authority.”
— Asia Today reported. Translation by UPI
© Asia Today. Unauthorized reproduction and redistribution is prohibited.
Korea original report: https://www.asiatoday.co.kr/kn/view.php?key=20260609010003141
