The evolution of artificial intelligence (AI) technology is upending the long-standing calculus of cybersecurity. The asymmetric dynamic—where attackers need to find just a single vulnerability while defenders must block every possibility—is poised to shift toward a defender advantage over the long term, as AI now enables defenders to actually examine all those possibilities.
Taesoo Kim, Microsoft Corporate Vice President of Security, made these remarks in an interview following his keynote address at Smart Cloud Show 2026, held at The Westin Josun Seoul in Sogong-dong on the 26th. Kim, who is also a professor of computer science at Georgia Tech, said, “In the AI era, offense and defense are not fundamentally different tasks. However, attackers are currently adopting AI rapidly, while defenders must contend with regulations, compliance requirements, and costs—so the offense currently holds the upper hand.”
He continued, “Defenders can decide whether code gets released, giving them more information than attackers. As AI advancement enables preemptive inspection of all possibilities, defenders will actually gain the advantage over the long term.”
Kim was appointed as an executive at Samsung Research in 2021 at the age of 36, where he led AI and cybersecurity research for four years. Last year, he led Team Atlanta to victory in the AI Cyber Challenge (AIxCC) hosted by the U.S. Defense Advanced Research Projects Agency (DARPA). AIxCC is a competition testing the ability to automatically analyze, detect, and fix vulnerabilities using AI. Team Atlanta scored 392.76 points in the finals, opening an overwhelming gap over the second-place finisher.
After moving to Microsoft as Corporate Vice President of Security earlier this year, Kim has led the development of MDASH, an AI-based vulnerability detection system. MDASH’s core approach combines different AI agents according to task stage and role. It builds threat models based on software architecture and historical vulnerability data, then deploys more than 100 sub-agents to hunt for vulnerabilities. Discovered vulnerabilities are cross-validated by agents playing the roles of hacker, developer, and defender, which also generate proof-of-concept (PoC) code confirming real-world exploitability and produce fix patches.
Kim explained, “If you feed the AI model that found a vulnerability back into the verification and discussion process, it may repeat similar conclusions due to the same biases. To prevent this, we apply different AI models to the developer-role and attacker-role agents to reflect diverse perspectives.” He added, “If the agents cannot reach consensus by the end, the three models vote, and a vulnerability is reported only when two or more judge it to be an actual bug.”
MDASH is currently integrated into the CI/CD pipeline of Microsoft’s Windows organization and is mandatory during the development process. Within four months of deployment, it identified vulnerabilities equivalent to 66% of all vulnerabilities discovered in Windows last year. Kim said, “Even though Windows undergoes continuous penetration testing, MDASH keeps finding new bugs in the same code. It is being recognized for discovering complex vulnerabilities at a level beyond human hackers in a short period of time.”
The Current State of South Korea’s Security Industry
Kim assessed that while South Korean security talent is world-class, the industrial base to support it has not yet grown sufficiently. In fact, Team Atlanta, which he led to victory at AIxCC last year, included researchers from Georgia Tech, KAIST, and POSTECH, with 80–90% of all members being Korean.
He noted, “South Korea is exposed to cyber threats from North Korea and other actors, so geopolitically it should treat cybersecurity as a priority, but the related industry is not sufficiently supporting this. The wage gap between U.S. and South Korean security professionals is severe, and domestically, security professionals often earn less than general software developers.”
He continued, “South Korean security talent leaving for higher compensation abroad reflects the fact that Korean society does not adequately value the security industry and its workforce. If the industry doesn’t change, government support alone could be like pouring water into a bottomless pit.”
How AI Is Reshaping the Future of the Security Workforce
Kim predicted that while AI will take over junior roles and narrow the traditional entry path for security professionals, new opportunities will also emerge. He said, “Even compared to six months ago, the pace of AI advancement is frightening. Since code—the foundation of security work—is text, AI’s reasoning capabilities are being applied rapidly, and technically AI already surpasses human hackers.”
However, he noted, “As AI-generated code increases, the scope of security work is actually expanding. Going forward, generalists who broadly understand AI, development, and security and can work flexibly across multiple domains will become increasingly important.” This suggests that rather than AI tools simply replacing repetitive tasks, human experts will be expected to bring broader vision and integrated judgment.
Click Here For The Original Source.
