Huang Renxun, CEO of Huida, pointed out that "cybersecurity" may become a large-scale application area under artificial intelligence (AI). AI accelerates the production of program code, and also accelerates the emergence, exploitation and repair of vulnerabilities, and the demand for "AI security" increases accordingly. The Presidential Office recently held the "International Forum on Defense Resilience of the Whole Society". Former US Deputy National Security Advisor Anne Neuberger suggested that Taiwan actively deploy "defensive AI". From industry to national security, the AI offensive and defensive competition has begun, and Taiwan’s information security defense line must be upgraded.
"Defensive AI" refers to the use of AI to assist in checking code, scanning vulnerabilities, identifying anomalies, integrating threat intelligence, and accelerating incident response. It uses machine speed to fight automated attacks, but major handling still requires manual review. When combined with existing information security controls, it can be regarded as a "digital immune system" for critical infrastructure, helping power, communications, medical and financial services maintain essential services. As far as the financial industry is concerned, AI may significantly shorten the time between the disclosure of vulnerabilities and the emergence of available attack methods. Financial institutions and their suppliers are therefore faced with the pressure of a large number of patches in a short period of time. This is not only a technical issue, but also a matter of continued operation of financial services and market trust.
Faced with China's long-term cyber intrusion, Taiwan is an important center for the global semiconductor and information communications supply chain. Once the defense line is lost, not only the operations of individual companies will be affected, but the global supply chain may also be affected. If it can combine chips, equipment, software and security talents, Taiwan can strengthen its resilience and develop internationally competitive AI security products and services.
The government should incorporate defensive AI into the security preparation of critical infrastructure, giving priority to energy, communications, medical and financial pilots, and establish a cross-departmental information integration mechanism and testing site. Models and tools should be tested under attack scenarios such as data poisoning and prompt word injection, and manual review, trajectory saving, deactivation and rollback mechanisms should be retained. The government can use digital twins to deduce chain attacks after network outages, power outages, or data tampering, and incorporate cyberattacks using agent-based AI into regular drills. In addition, secure backup communication channels must be established across agencies and key industry players to avoid coordination failures when a crisis occurs.
The industry should incorporate security requirements into software development and supply chain management, while the government can foster independent and trustworthy information security solutions through procurement, R&D subsidies, and demonstration sites. Defense should not just aim at "blocking", but should also adopt the mindset of "assuming it has been compromised" and implement zero-trust architecture, network segmentation, least privileges, offline backup and recovery drills. Externally, intelligence sharing and joint exercises should be deepened with the United States, Japan and European democratic partners. People should enable multi-factor authentication, not reuse passwords, check suspicious messages, and not click on unknown links.
Director of the Information Security Administration, Joseph Choi, attended an event recently and said that the Information Security Administration has incorporated AI into program vulnerability detection, intelligence analysis, offensive and defensive drills, etc., and will further expand its application next year. Taiwan has semiconductor strength, information security talent, and democratic resilience. The key lies in integration and execution. Only by building an "AI Shield" and implementing defensive AI into systems, tools, and daily drills can technological advantages be transformed into security advantages, strengthen Taiwan's digital resilience, safeguard the lifeblood of the economy and democracy, and transform security needs into new industrial competitiveness.
