Lecture on the blind spots of security in the AI era

Increase in system-bypass methods… Employee training is essential

[Reporter Lee Ji-wan, The Economist]

“The battle between spear and shield ​depends not on technology, but on time.”

Kim Jun-yeop, CEO of Lavarwave, spoke as a presenter at the 12th Economist Tech Forum held on the 24th at Community House Masil in Jung-gu, Seoul. His lecture focused on the theme “The real blind spots of security in the AI era.”

Attackers no longer directly attack advanced security systems. In the AI era, attackers steal and misuse valid credentials or obtain identities through phishing, accessing systems in the form of logins rather than intrusions. This is known as identity-based attacks.

Kim explained, “Even if you have good security products, you cannot feel safe. Attackers believe it is more effective to target company employees rather than directly attack strong security solutions.”

Direct, head-on malicious attacks still exist. However, identity-based attacks that bypass security systems have been increasing. According to the Microsoft Digital Defense Report, identity-based attacks increased by 32% in the first half of last year compared to the same period the previous year.

Kim added, “I once saw a video where attackers copied a user’s voice and issued commands without directly attacking Company A’s system. By acquiring identity, attackers make their actions appear as normal events. We must recognize that cases of neutralizing security systems in this way are increasing.”

Even Multi-Factor Authentication (MFA) cannot guarantee safety. Kim shared a personal experience:

“There have been repeated cases of spam emails sent to employees impersonating me. When I saw the emails, I was shocked at how similar the tone was to mine. That’s why I want to emphasize the importance of employee security training.”

He stressed that not only security staff but employees in all roles must receive thorough training.

“Internal education must come first so employees develop the habit of questioning events that appear as normal authenticated access from outside. The habit of constant skepticism is crucial.”

He continued, “Employees outside security roles often don’t fully grasp why breaches are dangerous. What matters is gaining their empathy. Companies must show how security incidents can reduce revenue and how such incidents could impact each employee’s role.”

Finally, Kim warned that AI—widely used by many—must also be handled carefully from an information security perspective.

“Today, AI itself can become a security gap. There is an attack method called AI poisoning, which infects code used by tools like ChatGPT or Gemini. Reference pages can be infected with malicious scripts, and once exposed, internal information can be transmitted as is.”