Amit Megiddo

The work of seeing and detecting is mature. The next chapter is enforcement, architecture, and secure-by-design. Native is now in AWS Security Hub Extended with multicloud coverage across AWS, Azure, GCP, and OCI.

AI agents are already being given real access to code, infrastructure, and production workflows. The winners will not be the organizations that slow that down. They will be the ones that enforce the right boundaries at the cloud level, so agents can move fast without putting the business at risk.

The threat is no longer theoretical. One low-skill actor used AI to compromise more than 600 devices in five weeks. This post explains why the real answer is not faster alerts, but security architecture that removes the conditions attackers are counting on.

AI adoption is accelerating, but most organizations are still governing it like software instead of architecture. This piece breaks down why model choice, data access, egress, and input handling are cloud security decisions, and what it takes to enforce the right boundaries before those decisions harden into risk.

Cloud providers offer powerful native controls, but operationalizing them across multiple clouds doesn't scale. Native is the control plane that translates security intent into enforcement at the source.