Why Microsoft calls AI agents 'insiders' #ai #cybersecurity #tech

AI agent security, Microsoft's version: Microsoft Defender lists Claude Code and every MCP server it uses on company devices. Microsoft now treats agents as insiders, and Agent 365 puts the controls outside the model. What it costs, and what you can copy. ---- Higgsfield AI — Create AI Videos & Visuals Get 10% OFF here: https://higgsfield.ai?fpr=thomas-dccf7e 🚀 DYNAMOUS AI COMMUNITY Want to learn agentic coding with live daily events and workshops? Check out Dynamous AI: https://dynamous.ai/?code=646a60 Get 10% off here 👉 https://shorturl.smartcode.diy/dynamous_ai_10_percent_discount ⚡ HOSTINGER — RELIABLE HOSTING FOR YOUR PROJECTS (10% OFF) Whether you're shipping a portfolio, a side project, n8n flows, or AI agents — I use Hostinger for fast, affordable VPS + web hosting. Get 10% off here 👉 https://hostinger.com/DIYSMARTCODE n8n: https://n8n.partnerlinks.io/diysmartcode (Affiliate links — costs you nothing, supports the channel.) ---- What you will see in this 2-minute breakdown: → Microsoft's Defender for Endpoint AI agent support matrix: 60 local agents, including Claude Code (agent and MCP servers), Cursor, Codex CLI, Ollama Desktop and LM Studio. Discovery is activity based: an agent shows up once Defender sees it running. → Microsoft's October 19, 2026 posts: agents as "agentic insiders", frequently over-privileged, and one prompt injection in an outside document can change what they do. → Satya Nadella's design rule: the controls sit outside the model. Dynamic Permissions starts an agent with no access and grants only what one task needs, only while that task runs, never more than the employee behind it. → Microsoft's expense-agent example: after an injected instruction its risk score climbs. It keeps read access, loses the right to approve, a human signs off the next big step, and it is only blocked at the top of the scale. → The price: Agent 365 is $15 per user per month (paid yearly) as an add-on to Microsoft 365 E5, A5 or Business Premium. Microsoft 365 E7 includes it at $99 per user per month. The new permission and risk controls are in private preview, with expanded availability promised at Microsoft Ignite (Nov 17-20, 2026). So far this is Microsoft's own description, not an independent test. → The pattern for your own agents: one key per agent with nothing attached by default, separate permissions for looking and changing, and the guard in a hook or proxy outside the model. Defender's runtime protection (preview, Windows) inspects Claude Code through its hooks. Microsoft Defender for Endpoint, AI agent support matrix: https://learn.microsoft.com/en-us/defender-endpoint/ai-agent-support-matrix Microsoft Security Blog, "Agents are your newest insiders" (Oct 19, 2026): https://www.microsoft.com/en-us/security/blog/2026/10/19/agents-are-your-newest-insiders-security-must-move-at-their-speed/ Official Microsoft Blog, "Securing the agentic insider" (Oct 19, 2026): https://blogs.microsoft.com/blog/2026/10/19/securing-the-agentic-insider/ Microsoft Agent 365 (pricing): https://www.microsoft.com/en-us/microsoft-agent-365 Microsoft Entra Agent ID (licensing): https://learn.microsoft.com/en-us/entra/agent-id/what-is-microsoft-entra-agent-id Your employer listing your Claude Code and its MCP servers: fair security, or a step too far? #AIAgentSecurity #ClaudeCode #ClaudeCodeSecurity #MicrosoftDefender #Agent365 #Microsoft365 #MicrosoftSecurity #PromptInjection #AgenticAI #AIAgents #AgenticAISecurity #MCP #MCPSecurity #AISecurity #Cybersecurity #AIGovernance #LeastPrivilege #ZeroTrust #EntraID #DevSecOps #DeveloperTools #DIYSmartCode