Skip to main content
0:00

A Practical Look at Insecure Agent Architectures

Joe Bollen, Snyk

Joe Bollen examines unsafe paths between agents and how to test whether the controls around them hold.

Vulnerabilities in multi-agent systems often emerge where agents interact. Joe Bollen draws on practical work attacking and defending these systems to explain how to find unsafe paths, contain them, and test the resulting controls.

format
Talk
topic
Agentic AI and autonomous systems
published