Security, Governance & Responsible AI

A first-class concern. This section covers identity, access management, data protection, compliance, governance, and responsible AI practices for agentic systems.

Agentic AI Security

📝 Secure Agentic AI End-to-End

Microsoft's end-to-end security model for AI agents — covers Agent 365, Entra, Defender, and Purview integration. Based on RSAC 2026 keynote content covering identity, posture management, threat protection, and data governance for autonomous agents.

Tags: Security · blog · architect, security · public · recommended · Agent 365, Entra, Defender, Purview · defense in depth · @JW-Sthlm · 2026-09-08

📝 Addressing OWASP Top 10 Risks in Agentic AI with Microsoft Copilot Studio

Maps each OWASP Agentic AI Top 10 risk to specific mitigations across Agent 365, Copilot Studio, Entra, Defender, and Purview. Use this to build your threat mitigation strategy for agent deployments.

Tags: Security · blog · architect, security, developer · public · recommended · Copilot Studio, Agent 365 · OWASP, threat mitigation · @JW-Sthlm · 2026-09-08

📖 Secure Autonomous Agentic AI Systems

Deep technical guidance on defense-in-depth for autonomous agents — model layer, safety systems, and runtime controls. Grounded in Zero Trust principles for securing fully autonomous AI workloads.

Tags: Security · documentation · architect, security · public · foundational · Zero Trust · defense in depth, autonomous agents · @JW-Sthlm · 2026-09-08

📖 Agent Identity Concepts in Microsoft Foundry

Explains agent identity blueprints, runtime identities, OAuth token exchange, RBAC, and downstream tool authentication in Microsoft Foundry. Use it to design least-privilege access without embedding credentials in agent code or prompts.

Tags: Security · documentation · architect, developer, security · public · recommended · Microsoft Foundry, Microsoft Entra Agent ID · agent identity, RBAC, tool authentication · @JW-Sthlm · 2026-09-08

📝 Least Privilege for AI Agents: Identity, Access, and Tool Binding

Practical guidance for treating each agent as a lifecycle-managed principal with explicit roles, tightly scoped permissions, and controlled tool access. The scenarios show how quiet permission growth and combined tool access can create risks that individual integrations hide.

Tags: Security · blog · architect, developer, security · public · recommended · Microsoft Entra, Microsoft Agent 365 · least privilege, tool binding · @JW-Sthlm · 2026-09-08

📝 Securing AI Agents as Tools Move from Reading to Acting

Microsoft Incident Response explains an MCP tool-poisoning attack against an agent with read-write access. It includes a practical playbook for detecting, containing, and preventing malicious tool metadata from triggering unauthorized actions.

Tags: Security · blog · architect, developer, security · public · advanced · Copilot Studio, Microsoft Foundry, MCP · tool poisoning, incident response · @JW-Sthlm · 2026-09-08

📝 Guarding AI Memory

Explains how persistent memory changes the agent threat model through delayed execution, memory poisoning, and reduced user awareness. It maps these risks to controls for memory creation, storage, observability, and user management.

Tags: Security · blog · architect, developer, security · public · recommended · Microsoft 365 Copilot · memory poisoning, delayed execution · @JW-Sthlm · 2026-09-08

📝 AutoJack: How a Web Page Can Reach the Host Running an AI Agent

Technical analysis of an AutoGen Studio exploit chain where untrusted web content reached an unauthenticated local MCP WebSocket and spawned processes. The broader design lesson applies to any browsing agent that can communicate with privileged localhost services.

Tags: Security · blog · architect, developer, security · public · advanced · AutoGen Studio, MCP · localhost trust, remote code execution · @JW-Sthlm · 2026-09-08

📝 Inside the Mastra npm Supply Chain Compromise

Detailed investigation of the compromise of more than 140 packages in the Mastra agent framework ecosystem. It covers the malicious postinstall chain, affected developer and CI environments, detection guidance, and package-supply-chain controls.

Tags: Security · blog · developer, security · public · advanced · Mastra, npm · software supply chain · @JW-Sthlm · 2026-09-08

📝 Building Secure, Governable AI Agents with Microsoft Foundry

How to build secure and governable AI agents using Microsoft Foundry — covers identity, access control, monitoring, and compliance patterns for agent development.

Tags: Security · blog · architect, developer, security · public · recommended · Microsoft Foundry · secure agent development · @JW-Sthlm · 2026-09-08

🏗️ Multi-Agent Reference Architecture — Security

Security chapter of Microsoft's multi-agent reference architecture — covers agent identity, inter-agent trust, data boundaries, and threat models for multi-agent systems.

Tags: Security · architecture · architect, security · public · advanced · — · multi-agent, security architecture · @JW-Sthlm · 2026-09-08

📖 Hosted agents in Foundry Agent Service - Microsoft Foundry

Explains how Microsoft Foundry hosted agents provide managed deployment and operation of AI agents, including the platform capabilities and considerations for secure agent workloads.

Tags: CAIP · documentation · architect, developer · public · recommended · Microsoft Foundry · managed agent hosting · @JW-Sthlm · 2026-09-16

📖 Hosted Agent Permissions Reference

Reference for the permissions required to create, deploy, and use hosted agents in Microsoft Foundry. Use it to assign least-privilege roles to deployment users and agent identities.

Tags: Security · documentation · architect, developer, security · public · recommended · Microsoft Foundry · hosted agents, RBAC, least privilege · @JW-Sthlm · 2026-09-16

Governance & Compliance

📖 Microsoft Agent 365 Documentation

Official product documentation for Microsoft Agent 365, the enterprise control plane for registering, observing, governing, and securing AI agents (Microsoft, open-source, and third-party) at scale.

Tags: Cross-solution · documentation · architect, security, delivery-lead · public · foundational · Microsoft Agent 365 · agent governance / control plane · @JW-Sthlm · 2026-06-17

📖 Protect Agent Identities with Microsoft Entra (Agent 365)

Microsoft Agent 365 admin guidance for protecting agent identities with Microsoft Entra — applying Conditional Access, lifecycle controls, and Zero Trust to agent identities governed through Agent 365.

Tags: Security · documentation · security, architect · public · recommended · Microsoft Agent 365, Microsoft Entra Agent ID · agent identity / Zero Trust · @JW-Sthlm · 2026-06-18

📖 Govern AI, Cloud Adoption Framework

Microsoft's Cloud Adoption Framework guidance for AI governance — roles, responsibilities, risk management, policy enforcement, compliance monitoring, and alignment with NIST AI RMF.

Tags: Cross-solution · documentation · architect, security, delivery-lead · partner-NDA · recommended · — · governance · @security-lead · 2026-09-08

📖 Azure AI Content Safety Documentation

Azure AI Content Safety service — detect and filter harmful content in text and images. Covers prompt shields, groundedness detection, protected material detection, custom categories, and content moderation APIs.

Tags: Security · documentation · architect, security, developer · partner-NDA · recommended · Microsoft Purview, Azure AI Content Safety · data protection · @security-lead · 2026-09-08

🏗️ Governance and Security for AI Agents Across the Organization

Cloud Adoption Framework reference architecture for agent governance — covers data, identity, observability, security, and development layers for enterprise-wide agent management.

Tags: Security · architecture · architect, security, delivery-lead · public · recommended · Cloud Adoption Framework · governance, agent lifecycle · @JW-Sthlm · 2026-09-08

📖 Secure and Govern Microsoft 365 Copilot Agents

Deployment model for securing and governing M365 Copilot agents using Microsoft Purview — covers data classification, DLP, and compliance controls for agent workloads.

Tags: Security · documentation · architect, security · public · recommended · Microsoft Purview, M365 Copilot · governance, data protection · @JW-Sthlm · 2026-09-08

📖 Prevent Data Leakage to Shadow AI

How to detect and prevent data leakage to unauthorized AI services using Microsoft Purview — DLP policies, endpoint controls, and monitoring to protect sensitive data from shadow AI usage.

Tags: Security · documentation · architect, security · public · recommended · Microsoft Purview · data loss prevention, shadow AI · @JW-Sthlm · 2026-09-08

📖 Copilot Studio Security and Governance

Official documentation on security and governance controls for Microsoft Copilot Studio — authentication, DLP, environment management, and audit logging for agent deployments.

Tags: Security · documentation · architect, security, developer · public · foundational · Copilot Studio · governance, security controls · @JW-Sthlm · 2026-09-08

📖 Introduction to the Agentic AI adoption maturity model

A framework for assessing and advancing an organization's adoption of AI agents, with repeatable practices for scaling agentic AI securely, measurably, and into everyday work.

Tags: CAIP · documentation · architect, developer · public · recommended · — · adoption maturity model · @frdeange · 2026-10-02

Responsible AI

📖 Govern Azure Platform Services for AI

Azure governance for AI platform services — policy enforcement, model governance, risk detection, Entra Agent ID integration, and operational standards for AI environments.

Tags: Cross-solution · documentation · architect, developer, security, delivery-lead · public · foundational · Azure AI Content Safety · responsible AI · @security-lead · 2026-09-08

📖 Responsible AI Policies

Cloud Adoption Framework guidance for establishing responsible AI policies across development, deployment, and operations. It provides a structured approach to governance, risk assessment, and accountability for AI agents.

Tags: Security · documentation · architect, security, delivery-lead · public · recommended · Cloud Adoption Framework · responsible AI, governance · @JW-Sthlm · 2026-09-08

Security Resources

📖 Azure AI Security Best Practices

Comprehensive security best practices for Azure AI services — covers authentication, network isolation, data protection, and monitoring for AI workloads.

Tags: Security · documentation · architect, security, developer · public · foundational · Azure AI Services · security best practices · @JW-Sthlm · 2026-09-08

📖 Security for AI Technical Documentation

Landing page for Microsoft's Security for AI documentation — covers threat protection, posture management, data security, and governance for AI workloads.

Tags: Security · documentation · architect, security · public · foundational · Security for AI · — · @JW-Sthlm · 2026-09-08

📖 Secure and Govern AI

Microsoft's central hub for securing and governing AI — links to guidance on threat protection, posture management, data security, and compliance across AI workloads.

Tags: Security · documentation · architect, security, seller · public · foundational · Security for AI · — · @JW-Sthlm · 2026-09-08

📝 Security for AI Announcements Blog

Latest announcements and updates on Microsoft's Security for AI capabilities — new features, product updates, and roadmap for securing AI workloads.

Tags: Security · blog · architect, security, seller · public · recommended · Security for AI · — · @JW-Sthlm · 2026-09-08