Azure AI Content Safety is a Microsoft cloud service that other companies can plug into their own products to automatically screen text and images for harmful content — such as hate speech, sexual material, violence, and self-harm. It launched in May 2023 and also powers safety filters inside Microsoft's own AI products, including Bing and GitHub Copilot.

Announced at Microsoft Build on May 23, 2023 and reached general availability later that year. The service uses machine learning models to evaluate content across four harm categories — hate, sexual, violence, and self-harm — and returns a severity score (0–6) per category. It supports eight or more languages and can be customized with company-specific blocked term lists. Azure AI Content Safety also includes Prompt Shields, which detect attempts to manipulate AI systems through crafted inputs (jailbreak attacks). It is a B2B (business-to-business) service, meaning businesses pay to use it via API rather than end users accessing it directly. TechCrunch noted that the launch came shortly after Microsoft dissolved its AI ethics team, raising questions about governance continuity.