TAIPEI (Taiwan News) — The Administration for Cyber Security said Thursday that emerging AI models are accelerating cyber threats, calling for stronger coordination in all sectors.
The agency said global cybersecurity researchers are tracking the offensive capabilities of advanced AI systems, per CNA. It cited reports that Anthropic’s Claude Mythos Preview identified thousands of high-severity vulnerabilities in widely used operating systems and browsers.
It also pointed to OpenAI’s GPT-5.5, saying the model significantly improves the efficiency and scale of cyberattack techniques. The agency warned that such tools are lowering barriers for attackers by making it easier to turn software flaws into usable exploits.
In response, the agency said it will adopt three measures: tracking defensive tools for emerging AI models, convening industry-government-academia consultations on cybersecurity strategy, and strengthening support for small and medium-sized enterprises through the Computer Emergency Response Team and Coordination Center.
It urged organizations to reinforce cybersecurity across strategy, management, and technology. At the strategic level, it said vulnerability management should be treated as an operational risk, with investment shifting toward resilience covering prevention, detection, response, and recovery.
At the management and technology levels, it recommended regular recovery drills, offline backups, and defense-in-depth systems. It also called for faster patching of vulnerabilities and disabling unnecessary external services to reduce attack surfaces.
