Microsoft teases agents that become 'independent users within the workforce' (www.theregister.com)

Microsoft has teased a new “class” of agentic AI it calls “independent users within the enterprise workforce” — essentially software agents with their own identity, dedicated access to organizational systems, and the ability to collaborate with humans and other agents. According to product roadmap documents and leaked admin screenshots, these agents (nicknamed “Agent 365” or A365) will have email and Teams accounts, entries in Entra ID/Azure AD and org charts, can attend meetings, edit documents, send/receive messages, and learn from interactions. Microsoft plans an M365 Agent Store and Teams discoverability; a targeted release is hinted for November around Ignite. For AI/ML teams and IT, the shift is technically notable and operationally risky. Giving agents persistent identities and access elevates needs for IAM, RBAC, DLP, audit logging, and model monitoring because agents can autonomously read, modify, and transmit enterprise data. Licensing and cost models add uncertainty: A365 appears separate from standard M365 and consumption pricing (Copilot credits) may make usage unpredictable. Security and governance questions — preventing rogue behavior, data exfiltration, incorrect or offensive outputs, and how to monitor/mitigate those behaviors — are unresolved. The offering could boost automation and productivity, but enterprise adoption will hinge on robust controls, clear pricing, and tooling for observability and policy enforcement.