Back Issues/Search Home → Calendar → Archive → RSS → Subscribe → Services → Vendors → Blog → Current Issue → Popular → Newsletters (LinkedIn, Substack)

All issues › Volume 344, Issue 2 › IT Vendor News › Fortinet

Beyond the Kill Switch: Containing AI Agents at Machine Speed

Fortinet, Wednesday, November 11th, 2026

Fortinet outlines how graduated containment, independent enforcement and human oversight can limit harm from misbehaving AI agents.

Fortinet describes its contributions to the World Economic Forum's Cyber Frontiers: AI and Cybersecurity initiative, focused on turning agent governance into enforceable controls.

Because AI agents can act across systems faster than humans can respond, it argues containment needs AI-enabled security that detects misuse and automatically restricts an agent's tools, credentials, context and memory.

Organizations should define each agent's identity, purpose, owner and authorized scope before deployment, then monitor whether activity stays consistent with its task.

Predefined criteria should set which actions to block and which need human review, while a kill switch remains a last resort and deception techniques such as canary credentials can signal scope violations.

more →  ·  More from Fortinet →