AWS expands AI asset inventory into Microsoft Azure

At a glance

AWS Security Hub AI Inventory can now discover self-hosted AI models, endpoints, and agents running on Azure virtual machines and correlate them with security findings. Why it matters: Multi-cloud organizations need visibility into where AI runs, which data it touches, and whether the underlying infrastructure is exposed.

What this means for Miami businesses

AWS says its Security Hub AI Inventory can extend visibility to AI models, endpoints and agents hosted on Microsoft Azure virtual machines. The announcement reflects a broader operational challenge: AI components can run across cloud boundaries, while security teams may already be working from separate inventories and alerting systems. For a South Florida organization using more than one cloud, the useful question is not simply “Where is the model?” It is also which account owns the workload, what identity can access it, what data flows through it, how it is exposed to the internet, and whether the configuration is monitored. Inventory is a starting point for answering those questions, not a substitute for remediation or governance.

Practical next steps

Map cloud accounts and subscriptions, identify owners, classify sensitive data, and review permissions and network exposure. Confirm which security products and service plans are required before treating any announced capability as available in your own environment. Cloud features change over time, and the right configuration depends on the customer’s tenancy and licensing.

How CG Networks can help

CG Networks supports cloud assessment, identity and security controls across Microsoft Azure and AWS environments.

Read the original update from Amazon Web Services. This article combines the linked source material with original analysis from CG Networks; verify current product availability and details with the vendor.

Talk with CG Networks

Back to the CG Networks Blog

Inventory is a beginning, not the whole security program

An AI inventory can help identify assets that might otherwise be overlooked, but an inventory only becomes useful when it connects to ownership and action. For each model, endpoint or agent, organizations need to know the responsible team, environment, identity, purpose, data access and exposure. Security teams should understand which alerts require review and how findings are routed to the people able to make configuration changes.

Cloud and AI services can involve account-level permissions and product prerequisites. Review the service’s current documentation and confirm that the feature is enabled and appropriate for the specific subscription before relying on it. Keep a separate record of accepted risks and remediation decisions. A discovery tool can improve visibility, but it does not itself set an organization’s access policy or resolve a finding.

Operational details worth checking

Organizations should distinguish discovery from response: a finding needs an accountable owner, a priority, a remediation or acceptance decision, and a follow-up check. Record exceptions with an expiry or review date where possible. This makes the inventory a living operational tool rather than a snapshot that becomes stale as cloud environments change.