Microsoft AI Red Team

About AI Red Team

Overview

• What is AI Red teaming and how Microsoft is building safer AI?

How-To Guide

• Guide for building AI Red Teams for LLMs

Reference

• Responsible AI tools and practices
• Responsible AI standard and impact assessment

Getting ready

Overview

• Microsoft's Open Automation Framework to Red Team Generative AI Systems (PyRIT)

How-To Guide

• PyRIT How to Guide

Reference

• AI Risk Assessment for ML Engineers
• AI shared responsibility model

Understanding Threats

How-To Guide

• Developer threat modeling guidance for ML systems

Concept

• Taxonomy for machine learning failure

Reference

• Bug Bar to triage attacks on ML systems

Exploring secure solutions

Concept

• Methodology for safety aligning the Phi-3 series of language models

Reference

• Enterprise security and governance for Azure Machine Learning
• What is Azure AI Content Safety?
• Harms mitigation strategies with Azure AI
• Monitor quality and safety of deployed prompt flow applications

Lessons learned

Overview

• Lessons from red teaming 100 generative AI products