Samsung SDS: "Proliferation of Generative AI Amplifies New Security Threats"

On February 23, Samsung SDS announced five major cybersecurity threats that companies should pay close attention to this year.

Yonhap News

Samsung SDS selected the following five items as the "Cybersecurity Threats in 2026": (1) AI-based security threats arising from the abuse and misuse of AI, (2) ransomware that takes control of user information and systems and demands payment in return for recovery, (3) cloud security threats that emerge as corporate IT environments are migrated to the cloud, (4) phishing and account takeover attacks in which attackers impersonate others to steal information or money, and (5) data security threats that cause unauthorized access to, damage to, or leakage of critical data.

In a survey conducted by Samsung SDS of 667 IT and security personnel in Korea, 81.2% of respondents cited AI-based security threats as their biggest concern. They pointed out that AI agents, which are evolving into autonomous actors in the workplace, could leak data or damage systems if they are given excessive delegation and authority. Ransomware (40.7%), cloud security threats (35.3%), phishing and account takeover (30.8%), and data security threats (26.6%) followed.

However, while more than half of the respondents are already using AI or plan to introduce it, 46.3% answered that they have no plans yet to introduce or apply AI for security purposes. Samsung SDS explained that this suggests that, despite increasing AI investment, proactive plans to address potential security threats remain insufficient.

As a countermeasure to AI-based security threats, Samsung SDS stated that it is necessary to apply an "AI guardrail" that monitors in real time the interactive process of inputting prompts into AI and receiving responses, as well as the outputs, in order to prevent information leakage. It added that, in the stages of building and operating AI systems, companies should conduct simulated attacks through an AI red team and remedy vulnerabilities to protect AI models and data.

Jang Yongmin, Head of the Security Business Team at Samsung SDS, said, "The spread of AI and AI agents will amplify new security threats, including sophisticated phishing, data leakage, and attacks targeting AI usage environments," and stressed, "We need to shift to a proactive posture by introducing AI-based security solutions that automate measures such as AI-based monitoring, detection, and automatic blocking."