detail
*It has been translated using AI
“Under Israel’s Cyber Dome system, its national, integrated artificial intelligence (AI) defense network, when one public institution detects a cyberattack, AI threat information is shared with every other institution within just minutes—not weeks later—and they all raise their defenses simultaneously. If one institution falls, the entire national payment network and confidence in the market are shaken. Korea, too, must establish a system in which the government, not just the financial sector, joins in defending against attacks at ‘machine speed.’”
As concerns over the threat of AI-driven hacking have surged, Dror Kashti, CEO of global cybersecurity firm Sweet Security (pictured), emphasized the need to build a national, real-time, collaborative AI defense network in an interview with Maeil Business Newspaper on the 25th.
Kashti served in the elite Unit 8200 of the Israel Defense Forces (IDF), regarded as one of the world’s leading intelligence organizations, and went on to serve as the IDF’s overall chief information security officer (CISO)—a reserve brigadier general overseeing about 2,200 cybersecurity experts. Toward the end of his military service, he oversaw security for Project Nimbus, a national strategic initiative to move Israeli government and military IT networks to private cloud infrastructure. He later founded Sweet Security, a company specializing in real-time cloud and AI security.
Kashti said that the recent incident in South Korea’s financial sector “is not an isolated problem but merely a ‘preview’ of the disaster that all financial systems around the world will face in the future,” identifying the essence of the incident as “the collapse of the barrier to entry for attacks” and “the reversal of speed.”
He noted, “In the past, hacking a single bank required a skilled team of professional hackers and enormous, nation-scale resources. Today, even ordinary people can use simple AI tools to find vulnerabilities in systems and break into them.” He added, “Until now, companies have managed security by finding vulnerabilities, prioritizing them, and then fixing them. But malicious AI exploits vulnerabilities far faster than security teams can fix them.”
He cited the “lag in adaptation” and “compliance for show” as structural reasons why advanced financial sectors, including Korea’s, are particularly vulnerable to AI-driven hacking. Kashti declared, “Financial firms handle their core operations on complex, outdated legacy systems built up over decades under strict regulations. That makes them easy targets for fast AI attacks,” adding, “Complying with legal regulations does not necessarily mean you are secure.”
Drawing on his past experience leading the Israeli military’s cloud migration project, he noted, “When financial firms rapidly adopt AI and cloud services, configuration errors and human-created gaps are inevitable.” He stressed the need to monitor the “runtime”—the point at which a service is live and running—in real time. In other words, rather than merely checking on paper whether a building’s blueprints show emergency exits, organizations need to install real-time CCTV inside the building and watch who is opening which doors and moving around. Kashti said, “The solution is to add another layer of security over existing IT networks: a new system that monitors operations in real time and finds and tests vulnerabilities in our systems before hackers do.”
