Forum Discussion

CrisDC's avatar
CrisDC
Copper Contributor
Oct 22, 2026

MS Purview DLP: Web search blocking also blocks Copilot response

Hello everyone,

I am currently testing a Microsoft Purview DLP policy for Microsoft 365 Copilot that is intended to prevent personal or sensitive information contained in a Copilot prompt from being used for a web search.

When a user enters a prompt containing specific Sensitive Information Types (SITs), only the web search should be blocked.

Copilot should still process the prompt and, if possible, provide a response without using web search or sending the detected sensitive information to the web search service.

Configuration

In Microsoft Purview, I only have the following actions available:

  1. Processing of prompts → Block
  2. Performing web searches → Block

I tested both options separately.

Behavior with “Processing prompts → Block”

This works as expected:

As soon as a configured SIT is detected in the prompt, processing is blocked immediately and Copilot displays a corresponding policy notification.

Behavior with “Performing web searches → Block”

I observe the following behavior:

  1. I send a prompt to Copilot that contains a SIT covered by the DLP policy.
  2. Copilot initially generates a normal response.
  3. In this response, Copilot indicates that the personal information was not used for the web search. However, the actual question is still answered.
  4. This response remains visible for approximately 1–2 seconds.
  5. The entire Copilot response then disappears.
  6. Instead, the following message is displayed:

“Due to your organization’s policies, none of the requested files or resources are available.”

The previously generated response, which was briefly visible, can no longer be accessed afterwards.

Expected behavior

Since I would like to use “Performing web searches → Block” in the DLP policy, I would expect that:

  • the web search is blocked,
  • sensitive information is not used for the web search,
  • Copilot can still process the prompt, and
  • the response generated by Copilot remains visible.
Question

Is this behavior expected when using the DLP action “Performing web searches → Block”?

In particular, I would like to understand why Copilot initially generates and displays a response without using the sensitive information for the web search, but then completely removes this response a few seconds later and replaces it with the policy notification.

Is it possible to configure additional actions in the DLP policy, or to select an option other than “Block”?

I only have the two actions mentioned above available.

I have attached the relevant screenshots. Please note that the attached screenshots are in German.

Thank you in advance!

Copilot response visible for 1-2 seconds

 

The response disappears and instead, this message displayed

 

The only two actions I can choose
copilot
dlp
dlp block
microsoft copilot
Microsoft Purview DLP
purview

4 Replies

  • Creativejodielawrance's avatar
    Creativejodielawrance
    Tin Contributor
    Oct 22, 2026

    Hello, this is what I believe the process to be: 

    When you submit a prompt, Copilot immediately starts generating its initial text response and querying external web sources concurrently. If the DLP policy detects that sensitive information triggered the web search block after the initial text stream has already started rendering on your screen, the safety/policy engine steps in immediately. It revokes and clears the generated output to prevent any potential data leakage to external search engines, replacing it with the policy notification block.

    • CrisDC's avatar
      CrisDC
      Copper Contributor
      Oct 22, 2026

      Yes, exactly. This is the behavior I am seeing.

      Do you know if there is a solution or workaround for this? Or is this a known issue with the DLP action “Block web searches”?

  • Jamony's avatar
    Jamony
    MCT
    Oct 22, 2026

    Your intended separation matches Microsoft’s documentation: “Performing Web Searches” blocks external web grounding when the configured SIT appears, while responses can still use permitted internal Microsoft 365 sources. “Processing prompts” is the separate action that prevents a response altogether.
    The briefly visible answer followed by a complete replacement is not explained by that documented distinction alone. Check every policy and rule applicable to the test user, including prompt-processing restrictions and sensitivity-label restrictions. Allow policy changes up to four hours to propagate before comparing results.
    Use synthetic sensitive data and a reproducible prompt, then compare the web and desktop experiences. Preserve the policy configuration, matched rule information, timestamps, and screenshots for a Microsoft support case.
    The documented action is Block for the chosen activity; there is no documented alternative here that guarantees the displayed answer will remain. Do not weaken the policy based on Copilot’s own claim that information was excluded.

    • CrisDC's avatar
      CrisDC
      Copper Contributor
      Oct 22, 2026

      Thanks. I have now also ruled out the other points:

      • More than 24 hours have passed since the policy configuration/change.
      • The behavior is identical in the Copilot desktop app and browser.
      • The two other DLP policies only apply to Teams and devices.
      • The Copilot DLP policy configuration has been verified.

      I have also reviewed the sensitivity label configuration and will verify that there is no additional Copilot DLP rule restricting labeled content.

      If this can also be ruled out, it seems that the observed behavior does not match the documented behavior of “Performing Web Searches”, since Copilot should block external web grounding but still be able to provide a response using permitted sources.

      Would you therefore consider this an issue that should be raised with Microsoft Support?