Error 555420: Resource blocked on Azure AI Foundry for Anthropic Claude model

LAURI HOWARD 25 Reputation points
2026-06-11T17:17:38.7233333+00:00

Hello Azure Support Team,

I recently deployed the Anthropic Claude model on my Azure AI Foundry resource named "gaskenbro-source". However, when testing the model in the Playground, I immediately encountered the following error: "ERR_BAD_REQUEST 555420: Your resource has been blocked because we detected unusual behavior."

This is a legitimate development environment for my security research and software development project, and no automated scraping or abusive requests have been made. It seems my resource was flagged incorrectly by the automated system.

Could you please review this case and unblock my resource so I can resume my project development? I have attached the screenshot of the error from the Playground.

Thank you for your assistance.

Foundry Models
Foundry Models

A catalog of AI models in Microsoft Foundry that you can discover, compare, and deploy using Azure’s built‑in tools for evaluation, fine‑tuning, and inference

2 answers

Sort by: Newest
Most helpful Newest Oldest
  1. Anshika Varshney 15,710 Reputation points Microsoft External Staff Moderator
    2026-06-15T17:40:50.5866667+00:00

    Hello @LAURI HOWARD

    We have received an update from the backend team that the deployment of the Fable 5 model is currently suspended due to a US government directive.

    For other models, please try again at a later time instead of making repeated requests.

    I Hope this helps. Do let me know if you have any further queries.

    If this answers your query, please do click Accept Answer and Yes for was this answer helpful.

    User's image

    Thank you.

    Was this answer helpful?

  2. AI answer

    2026-06-11T17:17:52.5566667+00:00

    Error 555420 indicates that the Azure AI Foundry resource has been blocked due to detected unusual behavior, which is governed by the Limited Access and abuse monitoring controls for advanced models like Anthropic Claude.

    To proceed:

    1. Review Limited Access and abuse monitoring
      • Check the Limited Access and abuse monitoring policies for Foundry models to ensure the intended security research and development use complies with the allowed scenarios.
      • If the project requires adjusting guardrails (content filters) or abuse monitoring, submit the appropriate registration forms:
        • Use the “Register to modify Guardrails (previously content filters)” form if guardrail changes are needed.
        • Use the “Register to modify abuse monitoring” form if monitoring behavior needs to be adjusted.
    2. Open a support request with Azure
      • Since the resource is already blocked, open a support ticket through the Azure portal under the Azure AI / Foundry resource.
      • In the ticket, include:
        • Subscription ID and resource name (for example, “gaskenbro-source”).
        • The exact error code and message: ERR_BAD_REQUEST 555420: Your resource has been blocked because we detected unusual behavior.
        • A short description of the workload (security research and software development), confirming that there is no automated scraping or abusive traffic.
      • Support can review the block, confirm whether it was triggered by abuse monitoring or other Limited Access controls, and advise whether the block can be lifted or if additional steps/approvals are required.
    3. Avoid repeated triggering while under review
      • Until support responds, avoid repeatedly sending requests from the Playground or code to the blocked resource, as this will continue to fail and might complicate the investigation.
    4. If using Claude under Limited Access
      • Some advanced models sold by Azure have stricter criteria for turning off or modifying abuse monitoring. The support team might require additional justification or may direct the use of the Limited Access registration flows before unblocking or adjusting the resource.

    If the resource remains blocked after support review, the support ticket is the channel to request further investigation or guidance on compliant usage patterns.

    References:

    AI-generated content may be incorrect. Read our transparency notes for more information.

    Was this answer helpful?

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.