Securing AI Agents in Hong Kong | Technical Workshop - Hong Kong

Secure AI Agents with Identity-First Controls

SYNOPSIS

JOIN OUR IDENTITY FOR AI WORKSHOP

AI agents are moving into real enterprise workflows, accessing apps, data, systems, and tools on behalf of users and teams. As adoption grows, identity and security leaders need a practical way to govern what these agents can access, what they can do, and when human approval is required.

At this workshop, you’ll learn how to reduce over-privileged AI access, move from impersonation to delegated access, apply runtime identity controls, and build human-in-the-loop approval paths for higher-risk actions.

Key Topics We’ll Cover

• Delegated access instead of impersonation - Move away from shared credentials and broad service accounts toward scoped, enterprise-managed delegated access.
• Runtime identity controls - Go beyond authentication and manage what AI agents can do after access is granted.
• Human-in-the-loop approvals - Add explicit human verification for high-risk actions, sensitive workflows, and critical decisions.
• Least privilege for AI agents - Reduce over-privileged access with short-lived tokens, time-bound scopes, and stronger governance.
• Identity propagation and accountability - Preserve visibility and accountability as activity moves across multi-agent workflows.