With the September release, AI security findings from Wiz flow directly into Unified Security Exposure Management (USEM) alongside every other exposure your team already works.

 

What you need to install

• Pre-requisite: You should have migrated your VR apps or plug-ins to use USEM (Unified Security Exposure Management).
• Install (or upgrade to) Vulnerability Response integration with Wiz version 32.8.4 or above.
• Install AI Security Exposure Management plug-in.
• Note: The AI Security Exposure Management plug-in is also available to any customer on a VR Enterprise license.

 

Today’s Wiz integration already brings vulnerabilities and misconfigurations across your cloud assets into USEM. This release extends that same pipe to AI.

 

Two kinds of AI security findings from Wiz

Wiz sends two distinct categories of AI risk into AI Security Exposure Management (AI SEM), and they arrive through two different paths.

 

 

 

1. Cloud configuration findings related to AI security

These come in through the existing Configuration Compliance integration. Think of a Bedrock agent deployed with no guardrails attached: a cloud configuration problem, but squarely an AI risk.

To route them, open the Wiz configuration, go to the Test Results Configuration tab, and select Send AI security findings to AI security exposure management.

 

 

 

With the setting on, Configuration Compliance test results that relate to AI security are rerouted to AI SEM and created as ' AI posture findings' instead of sitting in the general compliance queue. The advantage is that storing the AI security findings in the right tables within USEM will allow for visibility into AI security posture within AI control tower, mapping the findings to the right AI assets in CMDB rather than mapping them to generic cloud resource category, and access to any workflows that are specific to AI security. 

 

 

Note: This is opt-in today. From the December 2026 release , it becomes automatic for customers on USEM. This means, from December 2026, any findings related to AI security pulled from CC integration will automatically be routed to AI security exposure management tables.

 

2. AI security findings in code repos and hosted assets

This category of findings covers the risk that lives closer to the model and the agent itself. To ingest these findings, no additional configuration is required. If you already installed the latest Wiz VR integration plug-in (mentioned above) and enabled Wiz integration, the existing integration also fetches this new category of findings into USEM. No additional configuration required.

 

Vulnerabilities in open-source models:  Your teams pull models from public repositories, store them in an S3 bucket, retrain them, and ship them inside applications. Those model files can carry arbitrary code execution, unsafe operators, and pickle-based exploits that fire the moment the model loads. Most never get a CVE. They show up in USEM as AI vulnerabilities, tied to the specific model and the specific file.

 

 

 

Posture/configuration risks in agents and MCP servers:  Agents and MCP servers running in the cloud, or hosted on compute instances like EC2 and containers, get assessed too. A LangChain shell tool exposed where it shouldn’t be. An agent wired to a tool with no boundary around it. These land as AI posture findings against the hosted asset.

 

 

Every finding mapped to the right asset

A finding without context is just noise on a dashboard. So, every AI security finding from Wiz is mapped to the corresponding AI asset in your CMDB: the AI model, the agent, the MCP server, the tool.

That mapping is what makes the next step work.

AI Security Exposure Management ships an out-of-the-box assignment rule that identifies the business application mapped to the affected AI asset, then assigns the finding to that business application’s support group. Automatically.

 

 

What this means for your team

Your analysts stop juggling a separate AI security console and start working AI exposures in the same queue, with the ability to define risk calculators, assignment rules and other workflow automation features specific to AI security.

Also, AI stewards accessing AI control tower will get visibility into security metrics related to AI exposures and the remediation progress being made.

 

Install the plug-in to get visibility into and manage your AI attack surface within USEM.

 

Have you deployed Wiz to scan your AI assets for vulnerabilities or misconfigurations? Drop in your comments below.