[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-microsoft-adds-guardian-agents-to-police-other-ai-agents":10,"sections":41},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":30,"tags":31,"sources":36,"feedback":40,"feedback_at":22,"cost_usd":40,"total_tokens":40},11881,"microsoft-adds-guardian-agents-to-police-other-ai-agents","Microsoft Adds Guardian Agents to Police Other AI Agents","Microsoft's Guardian Agents supervise other AI agents and can restrict them as risk rises, treating software like a risky insider.","Microsoft is adding a layer of AI agents whose only job is to watch other AI agents, and it wants businesses to treat the watched agents as a new kind of insider threat.\n\nThe company calls the supervisors Guardian Agents and the risky workers \"agentic insiders\". In a blog post, Nadim Abdo, Microsoft's Corporate VP for Identity and Network Access, wrote that agents are not inherently malicious but can make mistakes or be manipulated, and that they need broad permissions to be useful. Abdo's central rule: \"We should not ask a model to be its own security authority.\" Controls over access and actions must sit outside the model, where it cannot bypass them.\n\nThe Guardian Agents extend Project Perception, Microsoft's system that coordinates red agents (probe for weaknesses), blue agents (investigate threats) and green agents (apply fixes). Microsoft also said it is expanding discovery to 35 more types of local agents, including Claude Code, OpenAI Codex and GitHub Copilot. Entra Agent ID, its identity system, gains identity types for local agents and for previously unknown agents that security sensors find.\n\nThree controls do the enforcing. Dynamic Permissions limit an agent to the access a specific task needs. Dynamic Agent Risk checks for threats. Dynamic Conditional Access can then restrict permissions, block certain operations, ask for human approval or, in extreme cases, stop and isolate the agent.\n\nMicrosoft's argument for using agents as the overseers is that they work at a speed humans cannot match. That is a vendor's framing, though it is not an unreasonable one.\n\nThe design borrows old ideas: least privilege and separation of duties, applied to software staff. Abdo describes surrounding non-deterministic intelligence (output that can vary run to run) with deterministic controls (rules that behave the same every time). How much of the Guardian layer falls into each camp is the useful test. Fixed rules are predictable. Model judgment inherits the same weaknesses the design is meant to contain.","[\"ai-agents\",\"microsoft\",\"enterprise-security\",\"identity\"]","2026-10-20T11:44:39.000Z","2026-10-20T12:08:39.642Z","2026-10-20T12:08:39.870Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"The closing paragraph asserts that Microsoft's announcement 'does not say' how much of the Guardian layer runs on fixed rules versus model judgment, but the writer only has TechRadar's secondhand summary, so that claim about the blog post's contents is unsupported and reads like a 'source says nothing' artifact; reframe the open question as the writer's own analysis without claiming what the announcement omits, and attribute the 'humans cannot match machine speed' point to Microsoft's framing as","resolved","security",[32,33,34,35],"ai-agents","microsoft","enterprise-security","identity",[37],{"name":38,"url":39},"TechRadar","https:\u002F\u002Fwww.techradar.com\u002Fpro\u002Fsecurity\u002Fmicrosoft-admits-frontier-models-cant-be-trusted-unveils-new-security-tools-to-treat-ai-agents-as-insider-threats",0,{"sections":42},[43,48,52,57,62,67,72,77,82,87,92,97,102,106],{"name":44,"slug":45,"count":46,"latest_published_at":47},"AI","ai",7042,"2026-10-20T16:00:00.000Z",{"name":49,"slug":30,"count":50,"latest_published_at":51},"Security",1015,"2026-10-20T12:00:00.000Z",{"name":53,"slug":54,"count":55,"latest_published_at":56},"Policy","policy",546,"2026-10-20T13:11:23.000Z",{"name":58,"slug":59,"count":60,"latest_published_at":61},"Deals","deals",536,"2026-10-20T15:55:20.000Z",{"name":63,"slug":64,"count":65,"latest_published_at":66},"Hardware","hardware",271,"2026-10-20T15:25:03.000Z",{"name":68,"slug":69,"count":70,"latest_published_at":71},"Consumer Tech","consumer-tech",252,"2026-10-20T16:04:13.000Z",{"name":73,"slug":74,"count":75,"latest_published_at":76},"Science","science",207,"2026-10-20T12:50:00.000Z",{"name":78,"slug":79,"count":80,"latest_published_at":81},"Startups","startups",135,"2026-10-20T11:00:00.000Z",{"name":83,"slug":84,"count":85,"latest_published_at":86},"Software","software",134,"2026-10-20T14:57:15.000Z",{"name":88,"slug":89,"count":90,"latest_published_at":91},"Dev Tools","dev-tools",119,"2026-10-20T13:00:00.000Z",{"name":93,"slug":94,"count":95,"latest_published_at":96},"General","general",88,"2026-10-20T09:36:16.000Z",{"name":98,"slug":99,"count":100,"latest_published_at":101},"Gaming","gaming",80,"2026-10-20T13:24:49.000Z",{"name":103,"slug":104,"count":105,"latest_published_at":51},"Reviews","reviews",38,{"name":107,"slug":108,"count":109,"latest_published_at":110},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]