Isolate Azure AD risky users

Azure Active Directory (AD) calculates a user’s risk level Open external link based on the probability that their account has been compromised. With Cloudflare Zero Trust, you can synchronize the Azure AD risky users list with Cloudflare Access and apply more stringent Zero Trust policies to users at higher risk.

This tutorial demonstrates how to automatically redirect users to a remote browser when they are deemed risky by Azure.

Time to complete:

1 hour

​​ Prerequisites

• Azure AD Premium P2 license
• Cloudflare Browser Isolation add-on
• Gateway HTTP filtering enabled on your devices
• (Recommended) wrangler installation

​​ 1. Set up Azure AD as an identity provider

Refer to our IdP setup instructions for Azure AD.

​​ 2. Add Azure AD API permissions

Once the base IdP integration is tested and working, enable additional permissions that will allow a script to create and update risky user groups in Azure AD:

1. In Azure Active Directory, go to App registrations.

2. Select the application you created for the IdP integration.

3. Go to API permissions and select Add a permission.

4. Select Microsoft Graph.

5. Select Application permissions and add the following permissions Open external link :

   • IdentityRiskyUser.ReadAll
   • Directory.ReadWriteAll
   • Group.Create
   • Group.ReadAll
   • GroupMember.ReadAll
   • GroupMember.ReadWriteAll

6. Select Grant admin consent.

You will see the list of enabled permissions.

​​ 3. Add risky users to Azure AD group

Next, configure an automated script that will populate an Azure AD security group with risky users.

To get started quickly, deploy our example Cloudflare Workers script by following the step-by-step instructions below. Alternatively, you can implement the script using Azure Functions Open external link or any other tool.

1. Authenticate wrangler.

   $ wrangler login
   

2. Open a terminal and clone our example project.

   $ wrangler generate risky-users https://github.com/cloudflare/msft-risky-user-ad-sync
   

3. Go to the project directory.

   $ cd risky-users
   

4. Modify wrangler.toml to include the following values:

   • <ACCOUNT_ID>: your Cloudflare account ID.
   • <TENANT_ID>: your Azure AD Directory (tenant) ID, obtained when setting up Azure AD as an identity provider.
   • <CLIENT_ID>: your Azure AD Application (client) ID, obtained when setting up Azure AD as an identity provider.

   wrangler.toml
   name = "risky-users"
   compatibility_date = "2023-01-04"
   main = "src/index.js"
   workers_dev = false
   
   account_id = "<ACCOUNT-ID>"
   
   [vars]
   AZURE_AD_TENANT_ID = "<TENANT-ID>"
   AZURE_AD_CLIENT_ID = "<CLIENT-ID>"
   
   [triggers]
   crons = ["* * * * *"]
   

5. Deploy the Worker to Cloudflare’s global network.

   $ npx wrangler deploy
   

6. Create a secret variable named AZURE_AD_CLIENT_SECRET.

   $ wrangler secret put AZURE_AD_CLIENT_SECRET
   

   You will be prompted to input the secret’s value. Enter the Client secret obtained when setting up AzureAD as an identity provider.

The Worker script will begin executing once per minute. To view realtime logs, run the following command and wait for the script to execute:

$ wrangler tail --format pretty

After the initial run, the auto-generated groups will appear in the Azure AD dashboard.

​​ 4. Synchronize risky user groups

Next, synchronize Azure AD risky user groups with Cloudflare Access:

1. Enable SCIM synchronization.

2. In Azure AD, assign the following groups to your SCIM enterprise application:

   • IdentityProtection-RiskyUser-RiskLevel-high
   • IdentityProtection-RiskyUser-RiskLevel-medium
   • IdentityProtection-RiskyUser-RiskLevel-low

Cloudflare Access will now synchronize changes in group membership with Azure AD. You can verify the synchronization status on the SCIM application’s Provisioning page.

​​ 5. Create a browser isolation policy

Finally, create a Gateway HTTP policy to isolate traffic for risky user groups.

1. In Zero Trust Open external link , go to Gateway > Firewall Policies > HTTP.

2. Select Add a policy.

3. Build an Isolate policy that contains a User Group Names rule. For example, the following policy serves app1.example.com and app2.example.com in a remote browser for all members flagged as high risk:

   Selector	Operator	Value	Logic	Action
   Domain	in	app1.example.com, app2.example.com	And	Isolate
   User Group Names	in	IdentityProtection-RiskyUser-RiskLevel-high

To test the policy, refer to the Microsoft documentation for simulating risky detections Open external link .