Guardrails in Microsoft Foundry: How Content Safety Scoring Works
Last Updated on October 29, 2026 by Editorial Team
Author(s): Katakamvivek
Originally published on Towards AI.
Severity scoring, input and output filtering, Python code and production error handling with Azure AI Content Safety
A guardrail is a check that sits between people and your model and decides, every single time, whether a message is safe to pass through. Without one, your chatbot or agent will eventually say something it shouldn’t, or do something it shouldn’t, and you will find out from a screenshot on social media.
After introducing why guardrails are essential, the article explains how to implement Azure AI Content Safety in Microsoft Foundry using two paths: no-code guardrails (portal-based) and code-based checks via the Content Safety API. It details what the system detects (harm categories and agent attacks like jailbreaks and prompt injection), how severity scoring works and how thresholds map to allowed vs blocked behavior, and describes the three-stage approach to filtering (input, middle/context for retrieved content/tool results, and output). The post provides practical Python examples for analyzing text, applying thresholds, handling Foundry blocks and API failures “fail closed,” and considerations for latency, cost, and production hardening. It also covers red teaming with Foundry’s evaluation agent, tuning guidance to manage false positives, and a closing checklist emphasizing scoring, three checkpoints, combining portal and code, logging, and proactive attacks.
Read the full blog for free on Medium.
Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.
Published via Towards AI
Towards AI Academy
We Build Enterprise-Grade AI. We'll Teach You to Master It Too.
15 engineers. 100,000+ students. Towards AI Academy teaches what actually survives production.
Start free — no commitment:
→ AI Engineering For Production — our new book, coming before the end of 2026. Join the waitlist.
→ Agents Architecture Cheatsheet — 3 years of architecture decisions in 6 pages
Our courses:
→ AI Engineering Certification — 90+ lessons from project selection to deployed product. The most comprehensive practical LLM course out there.
→ Agent Engineering Course — Hands on with production agent architectures, memory, routing, and eval frameworks — built from real enterprise engagements.
→ AI for Work — Understand, evaluate, and apply AI for complex work tasks.
Note: Article content contains the views of the contributing authors and not Towards AI.
Related posts
Recent Posts
![[Day 37/100] Build Your Own Agent Memory Layer with Postgres and pgvector](https://miro.medium.com/v2/resize:fit:700/1*-Doos2nmUT3FPZ-Qxbk3Hw.png "[Day 37/100] Build Your Own Agent Memory Layer with Postgres and pgvector")
