🔹 Feature: Azure AI Content Safety
🔹 What It Does: A comprehensive Azure AI service that helps you detect and reduce harmful content, including both user-generated and AI-generated material. It combines text and image moderation APIs with an interactive Content Safety Studio for testing, tuning, and operationalizing moderation strategies.
What it’s giving you:
✅ Text Moderation API: Detects hate, violence, sexual content, and self-harm with granular severity levels.
✅ Image Moderation API: Scans and flags images for the same harm categories with severity scoring.
✅ Prompt Shields: Protects LLM applications by scanning for prompt injection attacks and malicious user input.
✅ Groundedness Detection (Preview): Evaluates whether LLM responses are grounded in your source materials.
✅ Protected Material Text Detection: Flags AI output that matches protected text like lyrics, articles, recipes, or other selected web content.
✅ Task Adherence API: Identifies if AI agents are acting out of scope, prematurely, or misaligned in their workflows.
✅ Custom Categories (Preview): Define custom harmful-content categories for both text and images to quickly react to emerging risks.
