Using Microsegmentation to Contain Autonomous AI Agents
Akamai Technologies, Thursday, July 16th, 2026
Akamai argues perimeter controls fail against autonomous AI agents and that machine-speed microsegmentation is needed to stop lateral movement.
Jacob Abrams of Akamai explains why security models built around human-paced workflows break down when autonomous AI agents chain tools, memory, and other agents together.
Because agents operate at machine speed and blur trust boundaries, a single compromised agent can expand its blast radius across an environment far faster than defenders can respond.
The post positions microsegmentation as the containment layer that limits what any individual agent can reach, regardless of how it was subverted. Akamai frames this as breach survivability rather than breach prevention, accepting that some agent compromise is inevitable.