AI isn’t just making adversaries faster – it’s making them more effective. The recent Anthropic report on AI-orchestrated espionage showed how AI increased the attackers’ competence and context, all while operating at machine speed. Applying AI agents to automate reconnaissance, code generation, and lateral movement has broken the friction that has traditionally impacted each step of the kill chain.
For defenders, this new reality heightens a dangerous trade-off in today’s detection stacks: Signatures are too brittle to catch polymorphic AI code, and behavioral analytics generate too much noise when capturing "low-and-slow" agent activity that mimics legitimate administration. Security teams are stuck navigating a long list of "maybe this is a problem!" while the adversary iterates in seconds.
Join this session to learn how to eliminate this trade-off and regain the advantage. We will explore:
The Operational Shift: How AI agents automate the kill chain and why "Time-to-Pivot" is the new critical metric.
The Detection Gap: Why current tools struggle to distinguish competent agents from legitimate admins without unbearable noise.
The “No False Positives” Strategy: How to deploy high-fidelity Deception tactics including decoys, honeytokens, and breadcrumbs that convert attacker exploration into actionable intel for you.
Defending the AI Stack: New strategies for placing tripwires around your own AI infrastructure, spanning LLMs, pipelines, and agents, to catch sophisticated attacks early.
Product Manager
Zscaler
Principal Threat Researcher
Zscaler
Stop chasing probabilities. Learn how to force the adversary to reveal themselves with clarity.