Unable to deploy gpt-image-2 on Azure AI Foundry — error 715-123420, suspected RTFP block — requesting moderator help to create collaboration ticket

Randall Cabezas 0 Reputation points
2026-07-16T21:17:15.6866667+00:00

I was studying for a certification using Microsoft Learn Platform, In the exercises it has me create azure projects and then delete them. I am down to the last module and it asked to deploy gpt-image-2 model to the project I just created. When doing that I got the Error: "715-123420: Our system has detected this request as unusual activity for your account. If you are confident this is in error, please contact support."

I suspect that the constant removing of resources has caused a flag in the RTFP system. But I need to finish other courses and will have to find a way to not get blocked again.

DETAILS:

  • Azure subscription is: "Azure for Students"
  • Subscription Region is Locked to East US 2
  • Subscription ID: [Removed PII]
  • Resource group name: rg-rcabezas-ai-04
  • Model name(s) that failed: gpt-image-2
  • Correlation ID and timestamp from the failed deployment:
    • ID: 9d4e1ec5-9963-4c7c-8186-4c339bf72b42
    • Thursday, July 16, 2026, 09:07:26 PM (21:07:26 UTC) [approx]
    • Thu Jul 16 2026 15:07:26 GMT-0600 (Central Standard Time)
      • Note: "I cannot file a technical support ticket because I am on the Basic support plan I am a student"
Azure OpenAI in Foundry Models

1 answer

Sort by: Most helpful
  1. Dhruva Sharma 0 Reputation points
    2026-07-17T11:21:46.77+00:00

    Hi Randall,

    You've diagnosed this correctly. Error 715-123420 ("unusual activity for your account") is generated by Azure's Real-Time Fraud Protection (RTFP) system, not by anything wrong with your model deployment request itself. Repeatedly creating and deleting AI resources in a short window—exactly the pattern the Learn exercises push you through—is a known trigger. gpt-image-2 is also a higher-scrutiny model (image generation), so it gets flagged more aggressively than text models.

    Important: this is not something you can clear yourself. There is no config change, region switch, or retry that reliably lifts an RTFP flag. It has to be reviewed and released by the Azure account-protection team. That's why you're seeing the "contact support" message.

    What to do given you're on Azure for Students / Basic support:

    You do not need a paid support plan for this. Fraud/RTFP blocks fall under billing & account management, which is free on every subscription including Azure for Students. Go to the Azure portal → Help + support → Create a support request → for issue type choose Billing (not Technical). Describe it as an "account incorrectly flagged for unusual activity blocking model deployment." Billing tickets are free and this is the correct queue.

    Include exactly the details you already gathered, which is perfect:

    Subscription: Azure for Students, ID 8b112da2-9268-4403-ba42-3315dcedd0f9

    Region: East US 2

    Resource group: rg-rcabezas-ai-04

    Model: gpt-image-2

    Correlation ID: 9d4e1ec5-9963-4c7c-8186-4c339bf72b42

    Timestamp: 2026-07-16 21:07:26 UTC

    Explanation that this occurred during Microsoft Learn certification exercises that instruct repeated resource create/delete.

    Stop deleting/recreating resources for now. Additional churn while flagged can extend the hold. Wait for the review.

    To avoid re-triggering it on future courses:

    Reuse a single long-lived resource group and AI Foundry project across modules instead of deleting after each exercise. Delete only deployments within it, or just leave low-cost resources in place.

    Deploy models once and reuse them rather than redeploying the same model repeatedly.

    Space out create/delete operations rather than doing many in quick succession.

    On requesting moderator help for a "collaboration ticket": Moderators here can escalate an ICM/collaboration ticket to the product team, but for an RTFP/account flag the free Billing support request above is the fastest and correct path—it routes directly to the team that can lift the flag. If a moderator is available to escalate in parallel, providing the same correlation ID and subscription ID will help them.

    Hope you get unblocked quickly so you can finish your certification.

    Was this answer helpful?

    0 comments No comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.