How to Access Shadow AI Detection on CloudGen Firewall
This article explains how to enable Shadow AI Detection on a CloudGen Firewall (CGF) appliance, extending AI Security visibility and enforcement to on-premises and hybrid deployments alongside SecureEdge gateways. Shadow AI Detection is part of the AI Security feature in Barracuda SecureEdge. It identifies unauthorized AI-tool use across your network and lets you enforce allow or block policies for specific AI applications.
Shadow AI Detection is now available on CloudGen Firewall (CGF) appliances, extending the capabilities already on SecureEdge Edge Services to on‑premises and hybrid deployments. Policies are managed centrally in SecureEdge Manager and enforced across both platforms. CGF DNS telemetry feeds the AI Security dashboard in SecureEdge Manager and BarracudaONE, giving unified visibility into AI-tool usage across your network.
Why Shadow AI Detection Matters
Employees are increasingly adopting AI-powered tools such as chatbots, coding assistants, image generators, and productivity platforms without IT approval. This unauthorized use can expose sensitive corporate data, create compliance gaps, and introduce security risks that traditional web filtering solutions were not designed to address. Shadow AI Detection for CloudGen Firewall (CGF) helps organizations discover, monitor, and control the use of AI applications across their environment.
Key Benefits
Unified management - Configure Shadow AI policies once in SecureEdge Manager and enforce them across both SecureEdge Edge services and CloudGen Firewall appliances.
Improved visibility - Identify AI applications in use through CGF DNS telemetry, with insights available in the AI Security dashboard.
Policy enforcement - Allow or block specific AI applications across your existing CGF deployment.
Seamless convergence - Extend SecureEdge AI Security capabilities to CGF environments without replacing your existing firewall infrastructure.
Shadow AI Detection on CGF
Shadow AI Detection on CloudGen Firewall (CGF) uses two independent mechanisms: policy enforcement and DNS-based visibility. You can enable either capability separately.
Policy Enforcement
Shadow AI policies are configured in SecureEdge Manager and automatically synchronized to the CGF. The enforcement process works as follows:
AI Risk Services maintains a continuously updated catalog of AI applications and their associated domains.
In SecureEdge Manager, you define which AI applications to allow, block, or monitor.
SecureEdge Manager sends the scoped web-filter rules to the CGF through the existing secure management channel.
The CGF places the synchronized Shadow AI rules in a deterministic position within its URL filtering rule set and enforces them.
Synchronized Shadow AI rules are placed above the default URLCatDefault rule in the CGF URL filtering rule set, and this position is restored automatically on each synchronization. Manually reordering these rules in the CGF management interface is not persistent. Only Shadow AI-scoped web filter rules are synchronized, your existing manually configured firewall rules and web filter policies are not modified.
DNS Visibility and Reporting
When reporting is enabled, the CGF streams DNS telemetry to the SecureEdge Reporting Service. AI Risk Services analyzes and classifies the telemetry and makes the results available in:
The AI Security dashboard in SecureEdge Manager
BarracudaONE
DNS logging captures DNS queries from all clients behind the CloudGen Firewall, including both SecureEdge Access Agent-managed devices and LAN-connected devices. A dedicated origin type field in the dashboard distinguishes between agent-originated and LAN-originated queries. The CGF sends DNS-only telemetry for this feature
Requirements and Limitations
For AI shadow detection, use Barracuda CloudGen Firewall firmware version 10.5.1 or higher. Note: If you run CGF firmware 10.5.1, update it with HF1174. Download the hotfix from the Barracuda Networks download portal.
You must first enroll CGF in SecureEdge Manager via Integration > CloudGen Firewall. You can enroll standalone or Control Center–managed CGF appliances.
CGFs with an individual license are supported. Appliances licensed only via a pool are not eligible.
Following are the known issues and limitations:
Limitation | Details |
|---|---|
DoH/DoT bypass reduces DNS visibility | DNS-level visibility is reduced when clients bypass the CGF's local DNS cache by using DNS over HTTPS (DoH) or DNS over TLS (DoT). URL-filtering enforcement at the HTTPS layer remains active. |
Initial upload of historical DNS logs | When reporting is enabled for the first time, CloudGen Firewall may upload existing DNS log files. As a result, older records may appear in the AI Security dashboard, and the initial upload volume may be larger than normal. |
Mixed-hotfix Control Center deployments | In Control Center managed environments, all participating CloudGen Firewalls should run the same hotfix level. Appliances without the required hotfix may receive Shadow AI configuration but transmit more than the intended DNS-only telemetry. |
Workspace re-enrollment | After removing a CGF from one workspace and re-enrolling it in another, disable and re-enable reporting to ensure telemetry is routed to the new workspace. Historical telemetry remains associated with the original workspace. |
Disabling reporting removes dashboard visibility | Disabling reporting stops DNS log streaming to SecureEdge Manager. As a result, no new AI application activity appears in the AI Security dashboard. Shadow AI policy enforcement continues independently and remains active until the policies are disabled. |
Before You Begin
You must enroll the CGF in SecureEdge. If you have already enrolled the appliance in SecureEdge Manager, no additional steps are needed.
Confirm that your CGF serial number is registered. SecureEdge Reporting Service uses your serial number to map DNS telemetry to the correct account and workspace. If you have not registered it, reporting will not work correctly.
Note the expected synchronization timing:
After you make changes in SecureEdge Manager, allow approximately two minutes for policy synchronization to the CGF and approximately 7 to 10 minutes for activity to appear in the AI Security dashboard.
Step 1. Update the CGF to a Supported Firmware Version
Shadow AI Detection requires CGF firmware version 10.5.1 (with HF 1174) and higher. SecureEdge Manager will not push Shadow AI configuration to appliances running firmware earlier than 10.5.1. To check the installed firmware version, log in to the CGF management interface and review the system information. If an upgrade is required, install the supported firmware version before proceeding.
Control Center-managed deployments
All participating managed boxes must run the same hotfix level. In a mixed environment where the Control Center has the hotfix but a managed box does not that box may receive Shadow AI configuration while sending incorrect logs. Align all managed boxes to the same firmware before proceeding.
If your appliance already runs a supported firmware version, skip this step and proceed to Step 2.
Step 2. Enroll the CGF in SecureEdge
Shadow AI Detection requires the CGF to be enrolled in SecureEdge Manager. Enrollment also lets SecureEdge Manager push Shadow AI policies to the appliance.
If the appliance is already enrolled in SecureEdge Manager, proceed to Step 3.
To enroll a CGF that is not yet enrolled:
Go to https://se.barracudanetworks.com and log in with your existing Barracuda Cloud Control account.
In the top menu bar, click the Tenant/Workspace drop-down menu and select the workspace your appliance should be assigned to.
To retrieve the Registration Token from SecureEdge, go to Integration > CloudGen Firewall.
Generate registration token and paste the token into a text file.
Register an CGF appliance using the token.
Complete the enrollment and confirm that the appliance appears as connected in SecureEdge Manager before continuing.
For more information, see How to Configure a Barracuda CloudGen Firewall in Barracuda SecureEdge.
Step 3. Verify Reporting Status in SecureEdge Manager
Reporting must be enabled on the CloudGen Firewall (CGF) to stream DNS telemetry and populate the AI Security dashboard in SecureEdge Manager and BarracudaONE. Without reporting, Shadow AI policies continue to be enforced, but no activity data is displayed. To verify that reporting is enabled, go to Integration > CloudGen Firewall in SecureEdge Manager and confirm that the REPORTING column for your CGF appliance is set to Enabled. By default, Reporting is enabled. If reporting is disabled, see How to Enable Reporting on SecureEdge-Integrated CloudGen Firewalls.
Note
When reporting is enabled for the first time, the CloudGen Firewall (CGF) may upload existing DNS log data. As a result, historical activity can appear in the AI Security dashboard, and the initial data upload may be larger than expected.
Step 4. Configure Shadow AI Policies in SecureEdge Manager
Shadow AI policies define which AI applications are allowed, blocked, or monitored across your network. Policies can be applied by AI application category or by individual application. Policy changes are pushed to the CGF automatically, so you do not need to update the CGF directly.
SecureEdge AI Risk Services maintains a continuously updated catalog of AI applications, including:
Generative AI chatbots
AI coding assistants
AI image and video generators
AI-powered productivity tools
Enterprise AI platforms.
Note
Each application is assigned a risk category. Policies can be applied by category or to individual applications. User- and group-level policy exceptions are not available.
To configure shadow AI policies, follow these steps:
In SecureEdge Manager, go to AI Security > Overview.
Select Shadow AI.
For each AI application category or individual application you want to control, set the action to Allow, Block, or Monitor.
Save the policy. SecureEdge Manager pushes the scoped web-filter rules to the CGF through the existing secure management channel.
Synchronized rules are placed above the default URLCatDefault rule and restored to that position on each synchronization. Manual reordering of these rules is not persistent
Note
Only Shadow AI-scoped web-filter rules are synchronized. Your existing manually configured firewall rules and web filter policies are not modified.
Step 5. Verify the Configuration
After allowing approximately two minutes for policy synchronization, verify that Shadow AI Detection is active on the CGF.
Verify Policy Enforcement
Log in to the CGF management interface.
Open the URL filtering rule set and confirm that Shadow AI rules appear in the dedicated section above the default URL-category rule.
From a test client behind the CGF, attempt to access a domain that is blocked by your Shadow AI policy.
Confirm that the connection is denied.
Verify Dashboard Visibility
Allow approximately 7 to 10 minutes after generating test traffic.
In SecureEdge Manager, go to AI Security > Overview.
Confirm that the test activity appears in the AI Security dashboard.
Optionally, check BarracudaONE for the same activity record
(Optional) Disable Shadow AI Detection
You can disable enforcement, visibility, or both independently.
What to disable | Steps | Effect |
|---|---|---|
Enforcement only | In SecureEdge Manager, remove or disable the Shadow AI policies under AI Security > Overview > Shadow AI. The synchronized rules are removed from the CGF during the next policy synchronization. | AI application blocking stops, but DNS telemetry continues to be collected and remains visible in the AI Security dashboard. |
Visibility only | In SecureEdge Manager, disable reporting for the CGF box under Infrastructure > CloudGen Firewall > Reporting. | DNS log streaming stops and no new AI activity is displayed in the AI Security dashboard. Existing Shadow AI policies continue to be enforced. |
Enforcement and visibility (fully disabled) | Disable both the Shadow AI policies and CGF reporting. | Both enforcement and dashboard visibility are stopped. AI application blocking is disabled and DNS telemetry is no longer sent to SecureEdge Manager. No new AI activity is displayed in the AI Security dashboard. |
Note
Disabling reporting also disables the DNS telemetry required to populate the AI Security dashboard. As a result, no new AI application activity will be visible in the dashboard, even if Shadow AI policies remain enabled.
Further Information