Defining what’s important
Fairness
How might an AI system allocate opportunities, resources, and information in ways that are fair to the humans who use it?
Reliability and safety
How might a system function well for people across different use conditions and contexts, including ones that it wasn’t originally intended for?
Privacy and security
How might a system be designed to support privacy and security?
Inclusiveness
AI systems should empower everyone and engage all people, regardless of their backgrounds.
How might a system be designed to be inclusive for people of all abilities?
Transparency
How can we ensure that people correctly understand the capabilities of a system?
Accountability
People should be accountable for AI systems.
How can we create oversight so that humans can be accountable and in control?
The Microsoft Responsible AI Standard
Responsible AI in practice: Case studies from the 2026 RAI Transparency Report
- In fall 2025, Microsoft released memory in Foundry Agent Service, allowing developers to create and manage long-term memory for their agents. This case study shows how the team mapped, measured, and managed risks of Foundry Agent Service to ensure its safe release.
- Memory in Microsoft 365 Copilot enables the AI assistant to retain information about users across conversations to provide contextually relevant responses. This case highlights how the team navigated the Sensitive Use review process to develop and deploy capabilities like memory safely with high standards of privacy and security.
- Fara-7B is Microsoft’s first agentic small language model for computer use. This case highlights responsible AI governance at the model level, including evaluation, testing, and mitigation strategies before release.
- This case illustrates how Browse with Copilot, a feature enabling AI to perform web-based tasks, was developed responsibly by mapping prompt injection and privacy risks, red teaming realistic attacks, adding layered mitigations, and launching through preview with user controls and feedback loops.
Responsible AI Transparency Report
- Building trust at scale
- Meeting the agentic moment
- Lessons from deploying AI
- Investing in standards and tools
Transparency documents
Anomaly Detector
Azure AI Content Safety
Azure AI Face service
Azure AI Search
Azure Language in Foundry Tools
Azure OpenAI
Azure Video Indexer
Click to Do
Frequently asked questions
-
AI principles are guidelines designed to ensure the responsible development and deployment of AI technologies. These principles are crucial because they help mitigate risks, promote ethical practices, and maximize the benefits of AI for society.
The Responsible AI Standard at Microsoft consolidates essential practices to ensure compliance with emerging AI laws and regulations.
- Microsoft offers a range of tools and practices to help organizations practice responsible AI.
Additionally, the Responsible AI Standard at Microsoft helps define product development requirements for responsible AI. - Transparency Notes are created to help customers better understand the inner workings of AI technologies and make more informed decisions about their use. They are part of the Responsible AI Standard and are intended to support responsible AI development by providing insights into how AI systems are governed, mapped, measured, and managed.
Microsoft also offers the Responsible AI Transparency Report, which provides insights into how Microsoft builds apps with generative AI, oversees the deployment of those apps, supports customers as they build their own AI apps, and fosters a responsible AI community.
Follow Microsoft