The Information Security Administration fully introduces AI to strengthen defense, and vulnerability detection and attack and defense drills will be expanded next year

In the face of AI-accelerated information security attacks, the government has also accelerated efforts to improve information security defense. The Information Security Agency held a media exchange meeting today (5th). Information Security Agency Director Cai Fulong said that the Information Security Agency has currently introduced AI into program vulnerability detection, penetration testing, intelligence analysis, and offensive and defensive drills. Next year, it will further expand its application, including introducing AI penetration testing into the government's annual offensive and defensive drills, and plans to apply relevant results to transnational offensive and defensive drills.

In terms of program vulnerability detection, Cai Fulong pointed out that the current plan is to adopt a "4+1" model. In addition to testing four well-known foreign AI tools such as Microsoft and Google, it will also establish a local AI detection environment. Since AI tools from major international manufacturers may need to connect to the cloud, the Information Security Administration plans to establish local AI detection capabilities in the national network environment, which is expected to be completed by the end of this year.

After the local environment is completed, the systems of at least three government agencies will be tested initially, including common systems or central agency systems. Next year, it will be gradually extended to the core systems of other important government agencies, using AI to assist in program testing.

In addition to finding program loopholes, AI will also directly enter government offensive and defensive drills. Cai Fulong said that the government conducts offensive and defensive drills from April to October every year. In the past, penetration testing mainly relied on manual operations by engineers. Starting next year, AI tools will be added to use AI to conduct penetration testing and strengthen offensive and defensive drill capabilities.

In fact, this year, the “vulnerability hunting” red team has begun to use AI tools to mine vulnerabilities. Cai Fulong pointed out that when attackers have begun to use AI tools to find vulnerabilities, defenders can no longer rely solely on manual processing. Therefore, the Information Security Administration is gradually introducing AI elements into various information security tasks.

Another AI application locks in huge amounts of security information. Cai Fulong said that there are currently quite a lot of sources of intelligence collected by the Information Security Administration. If you rely solely on manual analysis, not only will the workload be huge, but it will also be difficult to process quickly.

Therefore, the Information Security Administration will also introduce AI for intelligence analysis next year. It is hoped that once important information security information is discovered in the future, government agencies or important key institutions can be notified more quickly, allowing relevant units to deploy and block threats as soon as possible. Relevant planning and research and development are expected to be completed by the end of next year.

Cai Fulong said that the cross-border offensive and defensive exercise "CODE (Cybersecurity Operations and Defense Exercise)" led by the Department of Digital Development and held every two years is also expected to actually introduce the offensive and defensive exercise structure and related practices constructed using AI next year (2027) as an important verification field for AI security defense capabilities.