Why Enterprises Aren't Moving the AI Attacks Everyone Can See Resource Card

The WSJ reported that Anthropic models hacked real companies during tests. Learn how last year's whiteboard is holding CISOs back from responding fast enough.

How CISOs Can Respond to the Latest Anthropic Model Breaches

Why Enterprises Aren't Moving on the AI Attacks Everyone Can See

A Containment Platform is a security architecture designed to limit how far an attacker can move once they get inside your cloud environment. Prevention and detection are still important, but they can't stop every breach. Containment ensures that when something does get through, the damage stays limited. It works by governing every communication path, enforcing policy across all workload types, and operating whether or not a threat has been detected. Aviatrix 9.0 strengthens this architecture by closing gaps that most security stacks still leave open.

What is a Containment Platform, and why does it matter?

The [Five Testable Properties](https://aviatrix.ai/platform/the-containment-platform/) are architectural requirements that define whether a platform can deliver real containment:

* __Path-complete__: Every communication path is governed with no bypasses.
* __Identity-aware at L7__: Policy is based on workload identity, not IP addresses.
* __Detection-independent__: Enforcement holds even if a breach hasn't been detected yet.
* __Compute-model agnostic__: VMs, containers, serverless functions, and AI agents are all covered without requiring agents on each workload.
* __Universally propagated__: A single policy reaches every cloud environment in subseconds.

These properties are measurable and can be tested against any environment.

What are the Five Testable Properties of a Containment Platform in cloud security?

Aviatrix 9.0 introduces several features that close common containment gaps:

* __DCF (Distributed Cloud Firewall) East-West with Transit Gateway__ adds enforcement to inter-VPC traffic without replacing your existing architecture.
* __Serverless DCF__ extends policy enforcement to Lambda, Azure Functions, and Cloud Run workloads with no agents or code changes.
* __DCF on Edge__ brings the same enforcement to edge compute environments.
* __IPv6 Phase 2__ ensures full dual-stack enforcement so IPv6 traffic can't bypass policy.
* __Kubernetes Discovery__ identifies K8s workloads natively without modifying clusters, CNIs, or pods.

What are the biggest new capabilities in Aviatrix 9.0?

Aviatrix 9.0 is designed around a "monitor before enforce" approach. Teams can roll out containment one VPC at a time, with built-in back-out states at every step. Features like Gateway Drain and Undrain let you perform maintenance without dropping active connections. Smart Gateways Phase 2 adds automated health monitoring and self-healing failover. These capabilities mean the enforcement plane stays stable during infrastructure changes, so security improvements don't come at the cost of application availability.

How does Aviatrix 9.0 avoid breaking applications during rollout?

Run the [Five Testable Properties](https://aviatrix.ai/platform/the-containment-platform/) against your current stack. Ask whether every communication path is governed, whether policy covers all workload types (including serverless and Kubernetes), whether enforcement depends on detecting a threat first, and whether policy propagates across all clouds in subseconds. Most organizations pass two or three of the five. 

After upgrading to 9.0, run the test again. The difference between your two scores represents your containment deficit. 

Your Aviatrix account team can [guide you](https://aviatrix.ai/contact-us/) through a structured upgrade path.

How do I know if my environment has containment gaps?

K8s. Serverless. East-West. Edge. IPv6. Now Contained.

The Aviatrix 9.0 release extends the Five Testable Properties of a containment platform with containment for Kubernetes, serverless, and all workload types.

Aviatrix 9.0: Containment for Kubernetes, Serverless, and All Workload Types

Zero Trust for AI Workloads

Aviatrix Zero Trust for AI Workloads delivers default-deny, network-layer Workload Containment for every AI agent and LLM proxy — no code changes, no SDK integration.

Extend Zero Trust Security to AI Workloads | Aviatrix

Zero Trust for AI Workloads: Network-Layer Containment for the Agentic Era 

144 to 1: The Security Blind Spot That Q1 2026 Just Proved 

Five supply chain attacks prove the Detection Era’s blind spot. Machine identities outnumber human identities 144 to 1 — and the Containment Era has begun.

After Five Supply Chain Attacks, The Containment Era Begins

144 to 1: The Security Blind Spot That Q1 2026 Just Proved

Aviatrix Zero Trust for Workloads Turning Frameworks into Runtime Control card image

Aviatrix Zero Trust for Workloads Turning Frameworks into Runtime Control

Learn how Zero Trust for Workloads protects workload-to-workload and internet-bound traffic with runtime security, microsegmentation, and continuous visibility across AWS, Azure, and Google Cloud.

Zero Trust for Workloads: Runtime Security and Enforcement Across Clouds

Aviatrix Zero Trust for Workloads: Turning Frameworks into Runtime Control

First Principles for Cloud & AI Security Why the Space Between Your Workloads Is the New Battleground card image

First Principles for Cloud & AI Security Why the Space Between Your Workloads Is the New Battleground

Preview Scott Leatherman’s InfoSec World session on zero trust, runtime threat protection, and First Principles for Cloud & AI Security. Learn more.

First Principles for Cloud & AI Security in Modern Networks

First Principles for Cloud & AI Security: Why the Space Between Your Workloads Is the New Battleground

The Trust Seam Why the Salesloft Breach Signals a New Era for Cloud Security card image

The Trust Seam Why the Salesloft Breach Signals a New Era for Cloud Security

The Salesloft Drift attack by UNC6395 reveals a critical network security gap. Learn how a Cloud Native Security Fabric (CNSF) can prevent similar attacks.

Willie Tejada

Author Biography

Published Articles by Willie Tejada

Why Enterprises Aren't Moving on the AI Attacks Everyone Can See

K8s. Serverless. East-West. Edge. IPv6. Now Contained.

Zero Trust for AI Workloads: Network-Layer Containment for the Agentic Era

144 to 1: The Security Blind Spot That Q1 2026 Just Proved

Aviatrix Zero Trust for Workloads: Turning Frameworks into Runtime Control

First Principles for Cloud & AI Security: Why the Space Between Your Workloads Is the New Battleground

The Trust Seam: Why the Salesloft Breach Signals a New Era for Cloud Security

Products

By Role

Customer Stories

Security & Legal

Solutions

Resources

Industry

Company

Tools