Microsoft is cracking down on behind‑the‑scenes apps that can override settings, install extra software or alter core Windows experiences without explicit user permission.

Introducing a “consent‑first model”, the tech firm said its new Windows Baseline Security Mode will give users full visibility of all apps and AI agents powering its products, and that any decisions these systems make will be reversible.

Answering customer calls for “stronger, more consistent security foundations”, Microsoft said that automated systems working within Windows will be restricted to clearly approved capabilities, with the company planning to collaborate alongside third‑party developers to ensure those limits.

At the heart of the new security model is a desire to improve user control over Windows, an ambition which Microsoft said means making software behaviour visible and permissions easier to understand.

With Baseline Security Mode, Windows will operate with runtime integrity protection switched on by default, preventing unauthorised changes to the OS and ensuring only trusted, signed software can run, though users and IT admins will be able to override these safeguards when needed. 

“Just like on your smartphone, Windows will now prompt you when apps try to access sensitive resources – like your files, camera or microphone – or when they attempt to install other unintended software,” said Logan Iyer, Windows Platform + Developer, writing on the company’s blog. 

“Apps and AI agents will also be expected to meet higher transparency standards, giving both users and IT administrators better visibility into their behaviours. These updates raise the bar for security and privacy on Windows, while giving you more control and confidence in how your system and data are accessed.”

Recommended reading

• Microsoft CEO Warns AI Risks Becoming a Speculative Bubble
• Report: 1.5 Million Enterprise AI Agents At Risk of ‘Going Rogue’
• Microsoft Confirms Global Price Hike for 365 Business Bundles

Microsoft said it is determined to preserve its tradition as an open platform, reaffirming its guiding principles of “freedom to install any app and openness to every developer”, with the firm already starting work with top-tier partners like Adobe, CrowdStrike, and OpenAI on a phased rollout of the new framework.

“As we build increasingly capable agents, it’s even more important for people to have visibility and control over what’s happening on their computers,” said Ari Weinstein, member of product staff at OpenAI.

“It’s great to see Microsoft level up the security of their platform, and we’re excited to work together to deliver powerful, secure AI experiences on Windows and beyond.”