Ai Risk Governance
The latest Ai Risk Governance coverage — news, analysis, and updates from the WindowsNews.AI desk.
Microsoft Turbocharges Excel Copilot With Finance Skills, Real-Time Data, and Change Tracking
Microsoft announced on June 25, 2026, that Copilot in Excel will gain reusable finance Skills, live data connectors to financial sources, and explicit change tracking. These additions aim to automate repetitive finance tasks, ensure data freshness, and build audit-ready spreadsheets, potentially reshaping how finance professionals use Excel.
Microsoft and BKW Harness AI Weather Models to Power Switzerland’s Green Energy Revolution
Microsoft has published a case study detailing its work with Swiss utility BKW to apply cloud-based AI weather models to renewable energy management. The collaboration uses Azure, Microsoft Research’s AI for Science, and advanced neural operators to deliver hyper-local, probabilistic forecasts that improve grid balancing, cut costs, and increase renewable penetration. The project demonstrates how enterprise geospatial intelligence can drive the energy transition.
MAI-Code-1-Flash Lands in GitHub Copilot: Agents Enforce Policy, Meters Run
Microsoft has made the agentic MAI-Code-1-Flash model generally available for GitHub Copilot Business and Enterprise as of June 26, 2026. Organizations must explicitly enable the model via admin policy, and all requests incur usage-based token costs, with admin-configurable budget caps to prevent overspending. The launch underscores a shift toward model governance and financial control as agentic coding capabilities reach commercial IDEs.
KARL STORZ Adopts Microsoft 365 Copilot as Governed AI for 10,000 Employees
On June 26, 2026, Microsoft published a customer story revealing how KARL STORZ, a German MedTech firm with nearly 10,000 employees, deployed Microsoft 365 Copilot under a rigorous governed AI infrastructure. The approach leveraged document hygiene, sensitivity labeling, and proactive compliance monitoring to protect sensitive medical data while empowering employees with generative AI. The case sets a new standard for safe AI adoption in regulated industries.
Microsoft Solves Multi-Account Identity Crisis in Outlook Calendar with New Copilot Account Picker
Microsoft has rolled out an account selector dropdown in the Copilot sidebar for Outlook Calendar on the web, letting users with multiple eligible accounts explicitly choose which calendar data and context the AI should use. The feature eliminates confusion when juggling work and personal Microsoft 365 subscriptions, preventing Copilot from inadvertently pulling information from the wrong account. The change is exclusive to the web client for now and requires no admin intervention.
Faster Data Breach Investigations: Microsoft Purview DSI's AI Analysis Tiers Arrive in March
Microsoft is introducing AI-driven analysis to Purview Data Security Investigations, with Standard and Advanced tiers. Public preview starts in March 2026, and general availability follows in April 2026, bringing automated event categorization and intelligent triage to security teams worldwide.
Microsoft Viva Agent Dashboard Gives IT Full Control Over Copilot Agent Creations Starting June 2026
Microsoft is launching a new Agent Dashboard Overview page in Microsoft Viva in June 2026, enabling IT administrators to track adoption, activity, and creators of custom AI agents built with Microsoft 365 Copilot. The dashboard provides real-time insights into agent usage, helps identify top creators, and strengthens governance for the growing agent ecosystem. This move underscores Microsoft's commitment to making AI agent management a central part of the employee experience.
Global and Viva Insights admins can now apply custom department-based filters to Copilot Power BI analytics starting June 2026
Microsoft has announced Roadmap Item 559995, a new governance feature allowing Global and Viva Insights admins to add custom organizational attribute filters to Copilot Power BI reports. Slated for June 2026, this capability gives enterprises finer control over who can view Copilot analytics data based on attributes like department or region, enhancing data security and compliance.
Microsoft Purview now auto-discovers all AI assets across Azure AI Foundry workspaces at the subscription level.
Microsoft has launched Purview integration for Azure AI Foundry, enabling subscription-level governance of AI assets across all workspaces. The feature automatically discovers models, datasets, and endpoints, provides full lineage tracking, and allows administrators to enforce compliance policies uniformly. This closes a major gap in enterprise AI governance and helps organizations meet regulatory demands without sacrificing developer agility.
Microsoft Edge’s New Tab Page to Become an AI Command Center: Copilot Integration Hits Preview Next Month
Microsoft has unveiled plans for an AI-infused new tab page in Edge for commercial customers, with a July 2026 preview and September 2026 general availability. The redesign centers on Copilot, offering intelligent assistance, enterprise governance, and a streamlined productivity experience.
Jensen Huang Slams Smuggled Nvidia AI Servers: 'National Security Risk and Technical Dead End'
Nvidia CEO Jensen Huang, at the company's 2026 annual stockholder meeting, warned that smuggled Nvidia AI servers are both a national-security threat and a technical dead end, leaving Windows IT environments vulnerable to unsupported hardware, driver failures, and potential firmware attacks.
The 2026 Custom Chip Land Grab: Inside OpenAI’s Jalapeño and Microsoft’s Master Plan to Tame AI Inference Costs
By 2026, OpenAI, Microsoft, and other tech giants will deploy custom AI chips like Jalapeño and Athena to slash inference costs and reduce reliance on Nvidia. The shift promises faster, cheaper AI for Windows users but introduces hardware fragmentation and software challenges.
Tecan Taps NVIDIA BioNeMo to Infuse Agentic AI Into Introspect Lab Platform
Tecan announced on June 24, 2026, that it is integrating NVIDIA’s BioNeMo Agent Toolkit into its Introspect lab analytics platform, enabling agentic AI for proactive, autonomous lab automation. The early access program begins in Q3 2026, targeting high-throughput labs with Windows-based deployment. This move promises to reduce downtime, accelerate root-cause analysis, and introduce new AI governance considerations for Windows IT teams in regulated industries.