How to fix Annotate and block indirect injection attacks alert for AI Foundry AzTB control on our subscription. Although we've already taken the necessary remediation steps, the KPI has not been resolved on the s360 board

Question

How to fix Annotate and block indirect injection attacks alert for AI Foundry AzTB control on our subscription. Although we've already taken the necessary remediation steps, the KPI has not been resolved on the s360 board

Niharika Punyamurthula 0 Microsoft External Staff

We received the Annotate and block indirect injection attacks alert for AI Foundry AzTB control on our subscription. Although we've already taken the necessary remediation steps, the KPI has not been resolved on the s360 board since December 19.

SRILAKSHMI C 16,975 Reputation points Microsoft External Staff Moderator

2026-01-09T14:11:21.21+00:00

Hi Niharika Punyamurthula,

I’m currently engaging with the backend team on this. Please allow us some time, and I will keep you updated.

Thank you!
Manas Mohanty 16,190 Reputation points Microsoft External Staff Moderator

2026-01-16T04:52:28.7266667+00:00
Hi Niharika Punyamurthula,

Thank you for sharing the relevant document in S360 alerts. You seem to be created content filter as per the remedial instructions and relay to GPT-4o deployment. We saw only one on UI.

Have requested to list all deployment and relay all the content filters. If there is no additional deployment except GPT-4o deployment, we should trigger the Powershell script from "Auto-remediate" to remediate s360 notifications.

Step -1: az login --identity , Step -2:az cognitiveservices account deployment list --name <resourcename> --resource-group <rgname>

Thank you for your inputs.

1 answer

Your answer

SRILAKSHMI C 16,975 Reputation points Microsoft External Staff Moderator

2026-01-09T14:11:21.21+00:00

Hi Niharika Punyamurthula,

I’m currently engaging with the backend team on this. Please allow us some time, and I will keep you updated.

Thank you!
Manas Mohanty 16,190 Reputation points Microsoft External Staff Moderator

2026-01-16T04:52:28.7266667+00:00

Hi Niharika Punyamurthula,

Thank you for sharing the relevant document in S360 alerts. You seem to be created content filter as per the remedial instructions and relay to GPT-4o deployment. We saw only one on UI.

Have requested to list all deployment and relay all the content filters. If there is no additional deployment except GPT-4o deployment, we should trigger the Powershell script from "Auto-remediate" to remediate s360 notifications.

Step -1: az login --identity , Step -2:az cognitiveservices account deployment list --name <resourcename> --resource-group <rgname>

Thank you for your inputs.

Answer 1

Hi Niharika Punyamurthula,

Thank you for confirming that s360 KPI alert mitigated for one of the resources and pending for other two.

Have suggested to

compare the content filter of remediated resources
Have suggested to use higher severity threshold for prompt and jail break shield.
Run the PowerShell script coming in auto-remediate incase the issue persisted.
Contact respective owner on KPI in office hours or send an email on steps followed and current status.

Please accept this answer as verified one if you found the inputs helpful.

Share via

How to fix Annotate and block indirect injection attacks alert for AI Foundry AzTB control on our subscription. Although we've already taken the necessary remediation steps, the KPI has not been resolved on the s360 board

1 answer

Your answer