Roadmap ID 577828 is titled "Microsoft Purview: Endpoint Data Loss Prevention — Extend Purview data security to the network layer through integration with Azure Firewall." Its status is In development. The listing targets public preview in November 2026 and general availability in March 2027 for the Worldwide (Standard Multi-Tenant) cloud.
Roadmap dates are targets, not promises. Microsoft's own roadmap page says all information is subject to change. As of this writing, Microsoft has not published Learn documentation for this specific Azure Firewall integration. A search for Azure Firewall-specific Purview DLP guidance turned up nothing official, so this article separates what the roadmap claims from what Microsoft already documents for related features.
What the roadmap item promises
According to the roadmap description, the integration would:
- Detect and protect sensitive data in supported network traffic, including text, files and AI interactions.
- Apply DLP policies to traffic from servers, virtual machines and other endpoints. That explicitly includes traffic from background services and automated workloads.
- Help stop unauthorized sharing with cloud applications through browsers, applications and APIs.
- Send activity signals to Microsoft Purview Insider Risk Management to help identify risky user behavior.
- Support alert management and investigation in both Microsoft Purview and Microsoft Defender.
The roadmap lists "Web" as the platform and names both Preview and General Availability release rings.
In short: Microsoft plans to use Azure Firewall as an enforcement point for Purview DLP. That would cover servers and automated workloads, not only users at managed PCs.
Why this changes Endpoint DLP's scope
The roadmap files this under Endpoint DLP, which is notable given how Microsoft currently defines that product. The Microsoft Purview service description says Endpoint DLP extends DLP to sensitive items stored on Windows 10, Windows 11 and macOS (Catalina 10.15 and later) devices. Policies are scoped to users who sign in to onboarded devices and are evaluated when a scoped user logs on.
That model needs two things: an onboarded device and a signed-in user. A Linux VM running a scheduled ETL job has neither. Neither does a Windows Server service account sending JSON to a third-party API at 3 a.m. Those workloads have mostly fallen outside Purview's user-centered coverage. The roadmap's mention of "background services and automated workloads" is aimed at that gap.
If the feature ships as described, the enforcement point moves from the device agent to the network path. Traffic that passes through Azure Firewall could be evaluated whatever generated it. That is my reading of the roadmap language, not a documented architecture. Microsoft has not said:
- which operating systems or workload types are covered
- whether an agent or client is required
- which Azure Firewall SKU is supported
- what network topology traffic must pass through
How Purview network data security works today
Purview already has a network-layer DLP capability, called network data security, but it works through other enforcement points. Microsoft Learn documents the following details.
Integrations. Purview network data security monitors, classifies and protects traffic to unmanaged and untrusted cloud apps, including generative AI apps. It works through Microsoft Entra Global Secure Access and through non-Microsoft SASE and secure-browser products. Partner integrations are listed in the Security Store inside Purview DLP. Azure Firewall is not currently documented as one of them.
Shared classifiers. It reuses the classifiers already configured in other Purview policies. Organizations can use collection policies for discovery, DLP policies for enforcement, or both.
Covered scenarios. Microsoft lists:
- AI interactions through browsers, apps and add-ins, such as ChatGPT, Gemini and Claude
- uploads to unsanctioned cloud storage
- email and attachments sent through cloud email providers such as Gmail
- online form submissions
- social media posts
Activities and actions. Four activities are supported: text sent to an app, text received from an app, files uploaded and files downloaded. DLP actions are Audit only and Block. Microsoft notes that supported activities and actions can vary by integration.
Protocols. All integrations classify HTTP and HTTPS traffic. WebSocket support varies by integration.
Limits.
- Policies do not apply to B2B guest users.
- Inline evaluation covers up to 4 MB of text and files up to 3 MB.
- After setup, allow up to 24 hours for policies to reach the network product and for the first data to appear.
Licensing. It depends on the integration:
| Integration | Documented licensing requirement |
|---|---|
| Entra Global Secure Access | Microsoft 365 E7 per-seat, or Purview E5 (or equivalent) plus Entra Internet Access (or equivalent) |
| Non-Microsoft SASE / secure browsers | Purview E5 (or equivalent) per-seat plus Purview pay-as-you-go billing |
| Azure Firewall (roadmap 577828) | Not disclosed |
For pay-as-you-go integrations, Microsoft bills per request, meaning each network call from a device or browser to a website or API. Responses are not counted. Do not assume any of these licensing or billing rules will carry over to the Azure Firewall integration.
In short: Purview can already enforce DLP on network traffic through Global Secure Access and partner SASE products. The new item adds Azure Firewall as an enforcement point and extends coverage to non-user workloads.
A preview of the setup: the Global Secure Access model
Microsoft has not published Azure Firewall configuration steps, but the Global Secure Access integration shows how network DLP is set up today. Microsoft's Entra documentation describes two halves:
- A network-side content policy. It uses the Scan with Purview action and selects the activities, content types and destinations to inspect.
- A matching Purview DLP policy. In the Purview portal you create an Inline web traffic policy, add cloud apps or adaptive app scopes, and enable Network and non-Microsoft secure browsers as the enforcement location. You then add conditions (sensitive information types or sensitivity labels) and set actions under Restrict browser and network activities. The policy can run in simulation mode before you turn it on.
Microsoft's documentation stresses that without a matching Purview DLP policy, the content policy cannot inspect content or enforce audit or block decisions. The actions in the DLP rule must also match the content types and activities chosen on the network side.
The Global Secure Access documentation also lists known limitations that are worth testing for in any network DLP deployment:
- Fail-open on inspection failure. If Purview cannot finish an inspection, for example because a payload is too large or an internal error occurs, the traffic is allowed.
- Identity required. Scan with Purview applies only to traffic tied to an Entra user identity. Unmapped traffic is not sent to Purview.
- Encoded files may escape inspection. Files embedded inside JSON or split across multiple encrypted requests may not be inspected.
- No OCR for traffic sent to Purview.
- QUIC (UDP) is not supported. Microsoft suggests a Windows Firewall rule that blocks outbound UDP 443 so browsers fall back to TCP.
The identity requirement matters most here. Global Secure Access inspects only traffic it can map to an Entra user. The Azure Firewall roadmap item promises coverage for automated workloads, which often have no user identity at all. How Microsoft attributes and enforces policy for identity-less server traffic is one of the most important unanswered questions for the preview.
Questions to ask when the preview arrives
Based on general Azure networking experience (this is analysis, not Microsoft guidance), admins should look for answers to the following.
- TLS inspection. Purview can't classify content it can't see, and most traffic to AI and SaaS services is encrypted. In Azure Firewall, TLS inspection has historically been a Premium-tier feature. Microsoft has not said which tier this integration needs.
- Routing. Workloads usually reach Azure Firewall through user-defined routes or a hub-and-spoke design. Any VM that bypasses the firewall, such as one with its own public IP or a misconfigured route table, would be outside enforcement.
- Fail-open or fail-closed. The Global Secure Access integration fails open on inspection errors. Production teams will want to know which behavior Azure Firewall uses before relying on it.
- Who an alert belongs to. The roadmap promises Insider Risk Management signals, which work best when activity maps to a person. When the traffic comes from a service principal or a shared VM, it is unclear who the "user" is.
- Cost. Network DLP through partner integrations is billed per request. Chatty automated workloads can generate a lot of requests.
Context and caveats
This fits a broader pattern of Purview extending into Azure infrastructure. Microsoft already documents an integration that shares Purview sensitivity data with Microsoft Defender for Cloud, which expands visibility into the data layer, enabling security teams to prioritize resources that contain sensitive data. Developers can also make AI agents policy-aware: one Microsoft MVP's write-up on the Purview SDK notes that with Purview SDK integration, your AI agents become policy-aware, enforcing DLP and access controls inline. An Azure Firewall integration would add a network-level backstop for workloads where no one has added the SDK.
There are reasons for caution, too:
- A roadmap entry is a statement of intent. It is not a product page, a price list or a deployment guide.
- Network DLP sits on the traffic path, so its limits can affect production. The documented 3 MB inspection cap and fail-open behavior in the existing integration are examples.
- Adding inspection to a firewall carrying server traffic could affect latency and throughput, and Microsoft has published no performance data.
What to do now
- Don't plan a production rollout around the March 2027 date. Treat it as a target.
- Pilot the existing network data security integration if you use Global Secure Access. The Purview side, including Inline web traffic DLP policies, simulation mode, and activity explorer filtered to the network enforcement plane, is likely to look similar.
- Inventory server and VM egress paths. Find out which workloads already route through Azure Firewall and which don't.
- Review your sensitive information types and labels. The integration reuses Purview classifiers, so weak classifiers will give weak results here too.
- Watch Microsoft Learn for Azure Firewall-specific prerequisites, supported SKUs, protocols, licensing and logging behavior as the preview starts.
The goal of covering automated workloads that make outbound connections without a user is sound. Whether Microsoft delivers it as described will depend on details it has not yet published.
References
- Microsoft Purview: Endpoint Data Loss Prevention-Extend Purview data security to the network layer through integration with Azure Firewall Microsoft 365 Roadmap · 2026-11-17T18:36:41.967573Z
- Learn about Microsoft Purview Network Data Security | Microsoft Learn learn.microsoft.com
- Create content policies for network content filtering - Global Secure Access | Microsoft Learn learn.microsoft.com