Build and secure systems that act on their own. This track covers agentic AI design, tool use and multi-agent orchestration alongside the new attack surface — and governance — that autonomy introduces.
Both how to build effective agents and how to keep them from becoming a liability.
Agentic systems don't just answer; they plan, call tools and take actions in the real world. That power brings a genuinely new class of risk: prompt injection, tool misuse, runaway loops and unclear accountability.
This track helps engineering and security teams design capable agents while keeping them controllable, observable and safe — with guardrails matched to the level of autonomy granted.
Agent architectures — planning, tool use, memory and multi-agent orchestration
The agentic attack surface — prompt injection, tool abuse and privilege escalation
Guardrails, sandboxing and human-in-the-loop controls for safe autonomy
Observability, evaluation and kill-switches for autonomous behavior
Governance and accountability for actions taken by autonomous systems
The Quantum Clock Is Ticking
Security experts estimate quantum computers capable of breaking RSA-2048 encryption could arrive by 2030-2035. Adversaries are already running "Harvest Now, Decrypt Later" campaigns. Upskilling your teams now is the difference between leading the transition and scrambling to catch up after the deadline.
Outcomes that let you adopt agentic AI without inheriting uncontrolled risk.
The ability to design capable agents with appropriate, bounded autonomy
A working threat model for agentic and multi-agent systems
Practical guardrail, sandbox and approval patterns to deploy safely
Monitoring and evaluation that catch unsafe behavior early
A governance approach that keeps accountability clear