Sysdig AI Defense Puts Runtime Guardrails on AI Agents
Help Net Security, Thursday, November 5th, 2026
Sysdig made AI Defense generally available to monitor AI agents at runtime and allow, hold or block risky actions across endpoint and cloud.
Sysdig announced general availability of Sysdig AI Defense, which monitors and controls AI agent activity across endpoints and cloud environments.
It checks agent actions against company policy to allow them, hold them for user approval or block them before completion, and records activity and decisions for investigations and audits.
Sysdig cites an IBM survey in which 77% of organizations said AI adoption is outpacing governance, and its JADEPUFFER research, where a ransomware agent fixed its own login failure in 31 seconds.
Early data shows the kernel sees about seven processes for every action an agent logs, so Sysdig argues teams need both agent logs and kernel-level visibility.