Azure AI Content Safety

service

proprietary managed Azure service(metered preview containers, cloud-connected)

features

text/image moderation, prompt shields, groundedness

source

closed

license

proprietary

Microsoft runs this as a proprietary Azure service and publishes no source code or model weights. Preview containers can run the text, image and Prompt Shields checks on customer hardware, but they are licensed only while connected to Azure for metering.

• https://azure.microsoft.com/en-us/products/ai-services/ai-content-safety recorded 2026-08-13

  Sold as an Azure service on consumption pricing - 'Flexible pricing to meet your needs', 'Pay for only what you use - no upfront costs' priced on analysis volume. The word self-host does not appear anywhere on the page, no source repository is linked, and use is governed by the Azure product terms rather than any published license.

• https://learn.microsoft.com/en-us/azure/ai-services/content-safety/how-to/containers/container-overview recorded 2026-09-29

  Microsoft's container guide says "Containers let you use a subset of the Azure AI Content Safety features in your own environment", covering analyze text, analyze image and Prompt Shields, notes that "The content safety container is available in public preview", and states that "Content safety containers aren't licensed to run without being connected to Azure for metering."

• https://learn.microsoft.com/en-us/azure/ai-services/content-safety/overview recorded 2026-09-29

  The service overview lists "An Azure subscription" among the prerequisites and says "All of these capabilities are handled by the Studio and its backend". Searched the body for open source, self-host and license: none, and the only GitHub links point to the documentation repository.

Microsoft positions this as the default content-safety layer in front of the Azure AI customer base, which is what the reading reflects. The company publishes no customer count, request volume or developer count for the service, so this is a credible claim about standing rather than a measured figure.

• https://azure.microsoft.com/en-us/products/ai-services/ai-content-safety recorded 2026-08-13

  Product and pricing page. No usage figure of any kind appears on it.

One service covers harm-category moderation for text and images, prompt-injection defense, groundedness checks, protected-material detection and a task-adherence check for AI agents. Groundedness, task adherence and custom categories are still in preview, and the protected-material, groundedness and custom-category models work in English only.

• https://azure.microsoft.com/en-us/products/ai-services/ai-content-safety recorded 2026-09-29

  The product page says "Identify and mitigate both direct and indirect threats with prompt shields" and "Pinpoint copyrighted content and provide sources for preexisting text and code with protected material detection". Searched the body for customer counts, request volumes and developer counts: none; the page names individual customer stories only.

• https://learn.microsoft.com/en-us/azure/ai-services/content-safety/overview recorded 2026-09-29

  The service overview lists Prompt Shields, which "Scans text for the risk of a User input attack on a Large Language Model", groundedness detection marked preview, protected material text detection, a "Task adherence API" that "Detects when tool use by AI agents is misaligned, unintended, or premature in the context of a user interaction", analyze text and image APIs that scan "for sexual content, violence, hate, and self harm with multi-severity levels", and custom categories in preview. It adds that the models for protected material, groundedness detection and custom categories "work with English only".