AI Usage

AI Usage shows the AI tools and applications Nebulock detects in your environment, which endpoints and users run them, and how that activity changes over time. Use it to find unapproved AI tools, see how widely each one is used, and start a hunt on any endpoint that needs a closer look.

AI Usage is under Insider Risk > AI Usage in the Nebulock portal.

Before you start

AI Usage requires the Insider Risk role. See User Roles and Permissions to assign it.

AI Usage detects tools from the telemetry Nebulock already collects. To see results, connect at least one of these data sources:

  • An EDR integration, such as CrowdStrike, SentinelOne, or Microsoft Defender
  • An identity provider, such as Okta or Microsoft Entra
  • AWS CloudTrail

See Connect Your Data Sources for setup steps.

Detected tools and applications

The AI Applications list shows every AI tool Nebulock has detected and how many endpoints each one appears on. Unapproved applications are listed first, shown in red and sorted by endpoint count. Approved applications appear at the bottom of the list in gray.

To change a tool's status, use AI Usage Settings.

Select an application to see its details. The summary cards at the top of the page update to show:

  • Endpoints: the number of endpoints where the application was detected
  • Percentage of endpoints: the share of your monitored endpoints that run the application
  • Events: the total number of events tied to the application

The Activity Timeline below the cards charts how many endpoints used the application each day over the past week.

Endpoints detected

The Endpoints Detected table lists every endpoint where the selected application was found, with the username, endpoint name, and event count for each one. Sort by Event Count to find the heaviest users.

Select a row to open Endpoint Details, which shows:

  • The users and host names associated with the endpoint
  • The endpoint ID
  • An activity timeline of the application's event count on that endpoint, by day

Hunt on an endpoint

To investigate an endpoint, select Hunt on its row. Nebulock starts a hunt with the detection details already filled in, so the agent begins with full context:

  • The application and endpoint, including the endpoint ID
  • An event summary: total events, time range, and the number of distinct endpoints and users
  • The top users on the endpoint
  • The top activity patterns, such as process command lines, process names, and DNS lookups

The hunt then investigates how the application is used on that endpoint and flags unusual volume or timing. Results appear in Hunts, where you can ask follow-up questions like any other hunt.

Endpoint activity

Below the endpoint details, the activity table lists what the application did on the selected endpoint, grouped into one-day buckets. Each row shows the time, user, category, action, tags, event count, and a sample path.

The labels above the table summarize what needs attention:

  • Actionable rows: the number of activity rows worth reviewing
  • Risky on this page: the number of rows on the current page that carry a risk tag

Risk tags flag activity that deserves a closer look. For example, contains_credential marks activity that may involve credentials or secrets.

Approve or unapprove AI applications

Mark each AI application as approved or unapproved for your organization. Your choices apply across AI Usage reporting.

  1. In the Nebulock portal, go to Settings > AI Usage Settings. You can also select the AI Usage Settings link at the top of the AI Usage page.
  2. Find the application in the AI Applications list. The list includes every AI application Nebulock can detect, including ones with no detections in your environment yet.
  3. Turn on the Approved toggle to approve the application, or turn it off to mark it as unapproved. Approved applications move to the bottom of the list on the AI Usage page and are shown in gray.

Use the Approval filter to show all applications, or only approved or unapproved ones. The Endpoint Total column shows how many endpoints each application appears on, so you can see the impact before you change its status.

Related


Did this page help you?