As artificial intelligence (AI) accelerates the pace of cyber attacks, South Korean companies face significant challenges in their security response systems. A recent Cisco survey revealed that South Korea's cyber defense capabilities fall below the global average, hindered by a lack of collaboration among security, network, and IT teams, as well as fragmented data.
On October 13, Cisco held a media briefing at the ASEM Tower in Gangnam, Seoul, to discuss 'Cybersecurity in the Age of AI' and released its 'Relentless Defense Report.' The term 'Relentless Defender' refers to organizations with high cyber defense capabilities.
Jatin Sachdeva, Cisco's Chief Security Architect for the Asia Pacific, Japan, and China (APJC), emphasized the need to strengthen fundamental security measures such as zero trust, network segmentation, and access management while leveraging AI to enhance detection and response speeds. An analysis of recent cyber attack cases in South Korea indicated that while AI was used to increase attack speeds, the actual breaches exploited known vulnerabilities and common attack vectors.
Hwang Seong-kyu, Cisco Korea's Security Business Executive, stated, "South Korea's cyber defense capability scored 58, below the global average of 64," adding that the proportion of Relentless Defenders in South Korea is only 1%, compared to 8% globally. The report surveyed 8,000 security professionals across 30 countries.
The survey also highlighted limitations in collaboration and data utilization among South Korean companies. Only 24% of respondents reported that their security, network, and IT teams operate as a cohesive unit, compared to 41% globally.
Furthermore, 44% of South Korean respondents indicated they spend more time collecting and connecting scattered data across various systems than on threat detection and response. Cisco advocates for improved collaboration and data accessibility to reduce response times.
With the rise of AI agents, managing access rights has emerged as a new security challenge. As AI agents increasingly access corporate data and applications directly, the need for stringent access controls has become critical.
Sachdeva stressed, "Users' permissions should not be directly granted to AI agents." He explained that organizations should assess the tasks performed by agents and the data they need to access, granting permissions only within necessary limits.
According to Cisco, while 54% of global respondents reported they could detect and respond to breaches involving AI agents, only 37% of South Korean respondents felt the same.
Cisco is also enhancing its cybersecurity capabilities in line with the rise of AI. The company plans to integrate the data analytics and observability capabilities of Splunk, which it acquired for approximately $28 billion in 2024, with its existing network and security operations to improve threat detection and response.
There is a growing recognition of the need to manage network and security data together rather than separately. As AI agents access data and applications through networks, analyzing endpoint information alongside network behavior is essential for identifying anomalies.
Additionally, Cisco introduced a cloud-based security solution, 'Cisco Secure Access,' which allows organizations to control the access and activities of AI agents. Cisco Secure Access is designed based on zero trust principles, managing user and AI agent access to applications and data while detecting and blocking unauthorized access or security policy violations.
Sachdeva noted, "By utilizing Cisco Secure Access, organizations can monitor the tools and data accessed by AI agents and detect and block any actions that exceed permitted boundaries."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.