Internet Access web content filtering
Microsoft's edit ·
Microsoft's commit message in MicrosoftDocs/entra-docs · commit ec162af · +108 −4 lines · also on GitHub
⋯ 1 unchanged line
22
title: How to configure Global Secure Access web content filtering
33
description: "Control internet access based on website categories, URLs, and FQDNs. Configure granular, user-aware filtering policies using security profiles and Conditional Access."
44
ms.topic: how-to
5
−ms.date: 04/18/2026
5
+ms.date: 05/29/2026
66
ms.subservice: entra-internet-access
77
ai-usage: ai-assisted
88
---
⋯ 8 unchanged lines
1717
1818
The web filtering feature currently supports user- and context-aware Uniform Resource Locator (URL)-based web category filtering, URL filtering, and FQDN filtering.
1919
20
+Web content filtering also supports two optional rule conditions that enable traffic-aware policy enforcement:
21
+
22
+- **Source traffic type filtering (preview)**: Scope rules to specific traffic types, such as agent, browser, or application.
23
+- **HTTP method request filtering (preview)**: Block or allow specific HTTP methods, such as GET, POST, PUT, PATCH, and DELETE.
24
+
2025
> [!TIP]
2126
> For file type-based filtering (MIME types) and integration with Microsoft Purview for data loss prevention, see [Create a content policy to filter network file content](how-to-network-content-filtering.md).
2227
⋯ 10 unchanged lines
3338
- User Datagram Protocol (UDP) traffic (that is, QUIC) isn't supported in the current preview of Internet Access. Most websites support fallback to Transmission Control Protocol (TCP) when QUIC can't be established. For an improved user experience, you can deploy a Windows Firewall rule that blocks outbound UDP 443: `New-NetFirewallRule -DisplayName "Block QUIC" -Direction Outbound -Action Block -Protocol UDP -RemotePort 443`.
3439
3540
41
+- To enforce source traffic type and HTTP method request rules for HTTPS traffic, enable TLS inspection. Without TLS inspection, only Server Name Indication (SNI)-based web content filtering rules apply.
42
+- Source traffic type filtering requires client-based Global Secure Access connections. Remote networks don't support source traffic type rules.
3643
- Review web content filtering concepts. For more information, see [web content filtering](concept-internet-access.md).
3744
3845
## High level steps
⋯ 21 unchanged lines
6067
- To match all subdomains of a domain, use the wildcard format `*.domain.com`. Note that the wildcard `*.domain.com` matches subdomains like `www.domain.com` but doesn't match the root domain `domain.com` itself. To cover both the domain and all its subdomains, include both entries as a comma-separated list (for example, `*.contoso.com,contoso.com`).
6168
- When entering multiple FQDNs in a comma-separated list, don't include spaces between entries (for example, `contoso.com,fabrikam.com,*.example.com`).
6269
- Note, the URL filtering Preview supports a maximum of 1,000 URLs per tenant.
70
+1. (Optional) Configure the **Source type** condition (preview). For more information, see [Configure source traffic type filtering (preview)](#configure-source-traffic-type-filtering-preview).
71
+1. (Optional) Configure the **HTTP method request** condition (preview). For more information, see [Configure HTTP method request filtering (preview)](#configure-http-method-request-filtering-preview).
6372
1. Select **Next** to review the policy and then select **Create policy**.
6473
74
+## Configure source traffic type filtering (preview)
75
+
76
+Source traffic type filtering lets you scope web content filtering rules to specific types of network traffic. You can enforce differentiated policies based on whether traffic originates from an AI agent, a web browser, or an application.
77
+
78
+### Supported source traffic types
79
+
80
+| Source type | Description |
81
+| --- | --- |
82
+| Agent | Traffic that originates from AI agents, such as Copilot agents or autonomous AI tools. |
83
+| Browser | Traffic that originates from web browsers. |
84
+| Application | Traffic that originates from desktop or mobile applications. |
85
+| Unknown | Traffic where the source type can't be determined. |
86
+
87
+### Configure the source traffic type condition
88
+
89
+1. In the web content filtering rule configuration, locate the **Source type** field.
90
+1. Enable the **Source type** field to include it in the rule.
91
+1. Select the source traffic types to include in the rule.
92
+
93
+> [!NOTE]
94
+> Source traffic type filtering is available only for client-based Global Secure Access connections. This capability depends on the Global Secure Access client sending task and processor metadata to classify traffic. Remote networks don't support source traffic type rules.
95
+
96
+> [!NOTE]
97
+> When a request's traffic type can't be determined, traffic is classified as **Unknown**. Traffic-type-specific rules don't match unless you explicitly target **Unknown** in a rule.
98
+
99
+### Example: Block AI agents from accessing social networking sites
100
+
101
+To prevent AI agents from accessing social networking websites while allowing browser and application traffic:
102
+
103
+1. Create a web content filtering policy rule.
104
+1. Select the **SocialNetworking** web category.
105
+1. Enable **Source type** and select **Agent**.
106
+1. Set the policy action to **Block**.
107
+
108
+This configuration blocks AI agent traffic to social networking sites while allowing browser and application users to access the same sites.
109
+
110
+## Configure HTTP method request filtering (preview)
111
+
112
+HTTP method request filtering lets you block or allow specific HTTP methods for matching traffic. You can enforce least-privilege access by restricting write operations while permitting read-only access.
113
+
114
+### Supported HTTP methods
115
+
116
+| Method | Description |
117
+| --- | --- |
118
+| GET | Retrieve a resource. |
119
+| POST | Submit data to a resource. |
120
+| PUT | Replace a resource. |
121
+| PATCH | Partially update a resource. |
122
+| DELETE | Remove a resource. |
123
+
124
+### Configure the HTTP method request condition
125
+
126
+1. In the web content filtering rule configuration, locate the **HTTP method request** field.
127
+1. Enable the **HTTP method request** field to include it in the rule.
128
+1. Select the HTTP methods to include in the rule.
129
+
130
+> [!IMPORTANT]
131
+> HTTP method enforcement requires TLS inspection for HTTPS traffic. Without TLS inspection, the Secure Web Gateway can't observe HTTP method headers, and only SNI-based web content filtering rules apply. Unencrypted HTTP traffic is always inspectable.
132
+
133
+### Example: Block AI agent write operations
134
+
135
+To prevent AI agents from performing write operations (PUT, PATCH, and DELETE) on enterprise resources:
136
+
137
+1. Create a web content filtering policy rule.
138
+1. Select the destination URLs, FQDNs, or web categories you want to protect.
139
+1. Enable **Source type** and select **Agent**.
140
+1. Enable **HTTP method request** and select **PUT**, **PATCH**, and **DELETE**.
141
+1. Set the policy action to **Block**.
142
+
143
+This configuration blocks AI agent traffic from performing write operations on the specified destinations while allowing GET and other read-only methods.
144
+
145
+## Combine source traffic type and HTTP method conditions
146
+
147
+You can use source traffic type and HTTP method request conditions independently or together in a single rule:
148
+
149
+- **Source type only**: Apply a rule to all traffic from a specific source type regardless of HTTP method.
150
+- **HTTP method request only**: Apply a rule to specific HTTP methods regardless of traffic source.
151
+- **Both conditions**: Apply a rule only when both the source traffic type and the HTTP method match.
152
+
153
+When you configure both conditions in a rule, a request must match both the specified source traffic type and the specified HTTP method for the rule to apply. Conditions are evaluated using AND logic across attributes.
154
+
155
+### Policy evaluation and precedence
156
+
157
+Policy evaluation follows the existing web content filtering policy framework ordering and precedence rules:
158
+
159
+- **Within the same attribute**: OR logic. Matching any selected value triggers the condition.
160
+- **Across attributes**: AND logic. All configured conditions must match.
161
+- **Conflicting actions**: When multiple policies match with conflicting actions, the most restrictive action wins (block over allow).
162
+
65163
## Create a security profile
66164
67165
Security profiles are a grouping of filtering policies. You can assign, or link, security profiles with Microsoft Entra Conditional Access policies. One security profile can contain multiple filtering policies. And one security profile can be associated with multiple Conditional Access policies.
⋯ 37 unchanged lines
105203
106204
You can use the baseline security profile to apply tenant-wide web content filtering policies to all remote network traffic. The baseline profile enforces policies at the lowest priority in the policy stack and applies to all Internet Access traffic routed through the service, making it ideal for securing remote network locations.
107205
206
+> [!IMPORTANT]
207
+> Source traffic type filtering isn't supported for remote network traffic. Source traffic type rules require the Global Secure Access client to provide task and processor metadata for traffic classification. You can still apply web category, URL, FQDN, and HTTP method rules without source traffic type conditions to remote networks through the baseline profile.
208
+
108209
### [Microsoft Entra admin center](#tab/microsoft-entra-admin-center)
109210
110211
1. Browse to **Global Secure Access** > **Secure** > **Security profiles** > **Baseline profile**.
⋯ 72 unchanged lines
183284
184285
|Step|Description|
185286
|-----|-----|
186
−|1|The Global Secure Access client attempts to connect to Microsoft's Security Service Edge solution.|
287
+|1|The Global Secure Access client attempts to connect to Microsoft's Security Service Edge solution. The client includes task and processor metadata for source traffic type classification.|
187288
|2|The client redirects to Microsoft Entra ID for authentication and authorization.|
188289
|3|The user and device authenticate. Authentication happens seamlessly when the user has a valid Primary Refresh Token (PRT).|
189290
|4|After the user and device authenticate, Conditional Access matches on Internet Access Conditional Access rules and adds applicable security profiles to the token. It enforces applicable authorization policies.|
190291
|5|Microsoft Entra ID presents the token to Microsoft Security Service Edge for validation.|
191292
|6|The tunnel establishes between the Global Secure Access client and Microsoft Security Service Edge.|
192293
|7|Traffic starts being acquired and tunnels through the Internet Access tunnel.|
193
−|8|Microsoft Security Service Edge evaluates the security policies in the access token in priority order. After it matches on a web content filtering rule, web content filtering policy evaluation stops.|
294
+|8|Microsoft Security Service Edge evaluates the security policies in the access token in priority order. It evaluates source traffic type, HTTP method, web category, URL, and FQDN conditions. After it matches on a web content filtering rule, web content filtering policy evaluation stops.|
194295
|9|Microsoft Security Service Edge enforces the security policies.|
195296
|10|Policy = block results in an error for HTTP traffic or a connection reset exception occurs for HTTPS traffic.|
196297
|11|Policy = allow results in traffic forwarding to the destination.|
⋯ 14 unchanged lines
211312
212313
1. Navigate to allowed and blocked sites and check if they behave properly. Browse to **Global Secure Access** > **Monitor** > **Traffic logs** to confirm traffic is blocked or allowed appropriately.
213314
315
+1. To verify source traffic type and HTTP method request rules, generate traffic from different source types. For example, use an AI agent to send a PATCH request to a protected resource, and confirm the rule enforces correctly in **Traffic logs**.
316
+
214317
The current blocking experience for all browsers includes a plaintext browser error for HTTP traffic and a "Connection Reset" browser error for HTTPS traffic.
215318
216319
⋯ 7 unchanged lines
224327
> [!NOTE]
225328
> To expedite Conditional Access configuration changes *for testing*, revoke user sessions in the Microsoft Entra admin center (select **Revoke sessions** on the user's overview page). This forces users to obtain new tokens with updated policies. Learn more about [Continuous Access Evaluation](concept-universal-continuous-access-evaluation.md).
226329
227
−## Next steps
330
+## Related content
228331
229332
- [Create a content policy to filter network file content](how-to-network-content-filtering.md)
230333
- [Learn about the traffic dashboard](concept-traffic-dashboard.md)
334
+- [Known limitations for Global Secure Access](reference-current-known-limitations.md)
Microsoft's Markdown source from MicrosoftDocs/entra-docs, © Microsoft Corporation, under MIT. Changed lines with up to 3 unchanged lines around each; ChangeIntel kept this copy 9 Oct 23:37 UTC. The commit date is when the source changed, which can be hours or days before Learn published it.