Agents Under Attack: Security Risks in the Era of Autonomous AI

Logo
Presented by

Shaked Reiner

About this talk

Agentic AI—systems powered by large language models (LLMs) that can act independently—are becoming increasingly common in modern workflows. From reviewing code to managing tickets and automation tasks, these systems introduce powerful capabilities and a radically new attack surface. This webinar breaks down how these agents work, what makes them vulnerable, and how attackers can exploit them using nothing but clever text. We’ll cover both the traditional attack surface and the new LLM-specific risks—and explore a demo agentic system to show these attacks live.
Related topics:

More from this channel

Upcoming talks (2)
On-demand talks (154)
Subscribers (22980)
Centered on intelligent privilege controls, CyberArk provides the most comprehensive security offering for any identity - seamlessly securing human and machine identities accessing workloads from hybrid to multi-cloud, and flexibly automates the identity lifecycle all with continuous threat detection and prevention – protecting organization’s identities and most critical assets by enabling Zero Trust and enforcing least privilege.