Discover AI Risk

This is a how-to guide about discovering AI risk via the Tenable One Cloud Exposure Console. For a conceptual overview of Tenable's AI-SPM capabilities, see AI-SPM.

Objective
Utilize Cloud Exposure AI-SPM capabilities to gain visibility into your AI/ML environment and manage unique AI-related risks (e.g., data poisoning, overprivileged agents).

Step 1: Discover AI Assets in Your Inventory

Use the Inventory to discover all AI resources (PaaS services, software, and data repositories) running in your environment.

  1. Navigate to Explore > Inventory.
  2. Expand the AI category.
  3. Search across cross-cloud AI resource types.

    The AI category in Inventory, listing AWS Bedrock custom models

Step 2: Correlate AI Resources to Training Data

One of the most critical steps in AI-SPM is establishing the relationship between an AI Model and the Sensitive Data it was trained on. This correlation is key to protecting against data poisoning and intellectual property theft.

  1. Discover the AI model: Start in the Inventory by locating your core AI asset. For example, an AWS Bedrock Custom Model.
  2. View resource details: Click on the specific Bedrock Custom Model name to open its resource profile page.
  3. Identify correlated data: On the Overview tab of the resource profile, you will find properties that link the model back to the source data. For example:

    • Training Input Data — Displays the specific S3 Bucket or repository used for fine-tuning the model.
    • Output Data — Displays the bucket used to store outputs generated by the model.

    An AWS Bedrock custom model's profile, with its training input and output data highlighted

  4. Assess data sensitivity: Click on the Training Input Data link. This takes you directly to the linked S3 Bucket resource page.
  5. From the bucket's resource page, navigate to the Data tab to confirm the presence of sensitive or proprietary information (e.g., PII, PCI) within the training dataset.