You can secure and monitor AI workloads that are deployed in private clouds, such as those built on ESXi and KVM servers. This capability extends protection to your AI applications and models even when they interact with public cloud Large Language Model (LLM) providers. By protecting the traffic between your private cloud workloads and external LLMs, you can safeguard against data exfiltration, prompt injection, and other threats specific to AI interactions. This functionality is essential for organizations with hybrid cloud strategies. It ensures that security is not a barrier to leveraging AI, allowing you to maintain control and visibility over your AI ecosystem regardless of where your data and applications are located.

To enable this, the Prisma AIRS™ AI Runtime: Network intercept can be manually deployed and bootstrapped in your private cloud environment. This deployment provides a crucial security layer for AI workloads that reside outside of public cloud infrastructure. Once deployed, the firewall can be centrally managed by either Strata™ Cloud Manager or Panorama, allowing for consistent policy enforcement and monitoring across your entire network.