AI workloads in production: what security teams need to control

TL;DR: AI security in production still lacks clear visibility, policy enforcement, and runtime control where models actually run, according to Aqua Security, even as 70% of AI applications are deployed in containers and 97% of security organisations plan to increase spending on securing AI use cases. The governance problem is no longer model experimentation; it is securing AI workloads as operational assets with enforceable controls.

NHIMG editorial — based on content published by Aqua Security: Operationalizing AI Security: Protecting Workloads Where AI Runs

By the numbers:

• Over 70% of AI applications are deployed in containers, running on Kubernetes and cloud native infrastructure.

Questions worth separating out

Q: How should security teams govern AI workloads running in production?

A: They should govern AI workloads at the runtime layer, where prompts, model outputs, and data access actually occur.

Q: Why do perimeter tools fall short for AI security?

A: Perimeter tools can see traffic, but they usually cannot see the model’s internal execution context or how policies are applied after the request enters the workload.

Q: What breaks when AI governance is limited to developer workflows?

A: What breaks is enforceability.

Practitioner guidance

• Map where AI actually runs Inventory every production location where inference, prompt handling, or model orchestration occurs, including containers and Kubernetes clusters.
• Separate edge controls from workload controls Document which policies are enforced by AI firewalls or SDKs and which are enforced inside the runtime environment.
• Align AI security with workload identity Use the same identity and access governance model you apply to machine workloads to define who or what can invoke models, pass prompts, or consume outputs in production.

What's in the full article

Aqua Security’s full article covers the operational detail this post intentionally leaves for the source:

• How Aqua positions workload-layer visibility and runtime protection for AI in production environments
• The Secure AI Advisory Program structure and what participants are expected to contribute
• Implementation detail on how the approach is integrated into the Aqua Platform for cloud-native environments
• Examples of runtime monitoring and policy enforcement across AI prompts, outputs, and usage patterns

👉 Read Aqua Security’s analysis of operationalizing AI security for production workloads →

AI workloads in production: what security teams need to control?

Explore further

View Full Forum →  |  NHI Foundation Course →