Park Sang-gyu (박상규), head of Palo Alto Networks Korea, urged companies to prepare on the assumption that a mythos-class AI could hack them and to pursue AI transformation and cybersecurity transformation at the same time. He called this year a turning point for AI hacking and cited the emergence of Anthropic Mythos and a hacking incident involving an OpenAI model and Hugging Face. He set out 10 principles for cybersecurity transformation, including frontier-AI testing, continuous readiness and SASE.

Cisco says security in the AI agent era requires integrating network and security to apply consistent policies and simplify troubleshooting. Mukesh Chand Tapriyal said AI agents change traffic patterns and increase connectivity demands while new risks such as shadow AI and prompt injection are rising. He outlined Cisco capabilities to identify AI apps and inspect open-source models. Cisco is accelerating its South Korea push, operating a Secure Access PoP since July.