Copilot Studio topic shows Responsible AI ‘ContentFiltered’ error on valid input

Question

Copilot Studio topic shows Responsible AI ‘ContentFiltered’ error on valid input

Anonymous

Hi everyone,

I was trying out copilot studio agents with topics. I’m getting the following error while testing my Copilot Studio agent that uses topics:

The content was filtered due to Responsible AI restrictions. Error code: ContentFiltered. Conversation ID: b430d9e2-bb56-4a5e-a688-c6168595af35 Time (UTC): 10/13/2025 6:51:59 AM

This happens when I enter “PI-2824” in response to a prompt asking for a policy number.

Interestingly, if I enter the exact same value (PI-2824) again after the error appears, the agent processes it correctly and moves on to the next step as expected.

It seems like the first attempt triggers some kind of false positive with Responsible AI content filtering, but the second attempt works fine.

Below is the screenshot of the transcript from the activity page: enter image description here

My topic flow is like this: enter image description here

Has anyone faced something similar or found a workaround? Would appreciate any insights on what might be causing this or how to prevent it.

Thanks, Satyabrata Samanta

Karan Shewale 2,585 Reputation points Microsoft External Staff

2025-10-13T09:23:05.6266667+00:00

Thanks for bringing this issue to our attention. We will check and update you soon.

1 answer

Your answer

Karan Shewale 2,585 Reputation points Microsoft External Staff

2025-10-13T09:23:05.6266667+00:00

Thanks for bringing this issue to our attention. We will check and update you soon.

Answer 1

Hi Satyabrata Samanta [UNext], this is a known issue in Copilot Studio where the Responsible AI filter may incorrectly flag certain inputs as problematic on the first attempt. In your case, the input “PI-2824” is being falsely identified as sensitive or restricted content.

To resolve this, you can make your prompt more specific — for example, ask users to “Please provide your policy reference (format: XX-#### like AB-1234).” You can also add input validation using a condition node to ensure the entered text matches the correct policy number format. Another effective solution is to create a custom entity for policy numbers instead of accepting free text input.

If the issue persists, report it to Microsoft Support with your conversation ID (b430d9e2-bb56-4a5e-a688-c6168595af35) and mention that “PI-2824” is a legitimate policy number. This will help Microsoft improve the accuracy of its content filtering system.

The reason the second attempt works is that the system learns from context and adjusts the filter behavior temporarily. However, implementing proper input formatting and validation will prevent such false positives from happening in the first place.Official Microsoft Documentation:

Thank you.

Karan Shewale.

If this response resolves your issue, please accept the answer and, if helpful, click the “Upvote” button. Your feedback helps us improve and assist others more effectively.