Microsoft Outlines Cyber Defenses for Agentic AI Era at Taiwan Cybersecurity Conference

Agentic AI can independently plan and execute multistep tasks, while also enabling cyberattacks to become more automated and scalable. Microsoft said companies that continue to handle alerts manually one by one will struggle to keep pace with threats and must use a Zero Trust architecture to unify protections across identities, devices, data and applications.

During the 2026 Taiwan cybersecurity conference, Microsoft Global Senior Vice President Vasu Jakkal said corporate cybersecurity should shift toward “ambient” defenses that continuously detect risks and “autonomous” defenses that respond automatically. Companies should also treat AI agents as high-risk identities and bring them under permissions and access governance. No investment amount was disclosed.