Microsoft Scout is an autonomous AI agent built into Microsoft 365 that can take actions on your behalf. This guide explains what it does, how it works, how it compares with other AI agents, and what developers need to know about AI agent permissions and security.
Microsoft Scout is rolling out as part of Microsoft 365 Copilot for enterprise customers. Availability varies by region and license tier. Check your Microsoft 365 admin center for Scout features under Copilot settings.
Microsoft Copilot answers questions and generates content ("what were last month's sales?"). Scout takes autonomous action ("update the Q3 forecast spreadsheet with these numbers and notify the team"). Scout is the action layer on top of Copilot's intelligence layer.
Scout operates within your existing Microsoft 365 permissions via Entra ID (Azure AD). It can only access data and apps you already have permission to use. For sensitive actions, Scout requires explicit human approval before executing.
Yes. Microsoft provides the Copilot extensibility platform, allowing developers to build custom agents using Teams AI Library, Azure AI Foundry, and the Microsoft Graph API. These agents can be scoped to specific tasks and datasets.
In June 2026, Meta confirmed that an AI-powered customer support chatbot was exploited to take over thousands of Instagram accounts. The chatbot had permissions to transfer account ownership — something users didn't expect from a support bot. This incident sparked the current focus on AI agent permission auditing.
Singapore launched the world's first government-mandated AI agent registry in 2026. Organizations must register autonomous AI agents and declare what systems they can access, what actions they can take, and who is responsible for their decisions.