Windows 11 adds consent checks for artificial intelligence agents accessing personal files

Microsoft is introducing a consent framework in Windows 11 that requires artificial intelligence agents to request explicit permission before accessing key personal folders, with controls applied individually to each assistant. Users will be able to decide how and when agents can reach files or system features, or block them entirely.

Microsoft has introduced a mandatory consent framework in Windows 11 that prevents artificial intelligence agents from accessing personal files without explicit user approval. In updated documentation for experimental agentic artificial intelligence features, the company specifies that six folders are designated as protected locations by default: Desktop, Documents, Downloads, Music, Pictures, and Videos. Each artificial intelligence assistant must request access to these locations individually, and the operating system does not grant any system-wide file permissions automatically.

The new controls operate on a per-agent basis, so approval granted to one artificial intelligence tool does not extend to other assistants that might be installed on the same system. When an agent attempts to reach files in protected folders, Windows displays a consent interface where users can choose whether to grant permanent access, require reauthorization every time the agent interacts with those files, or deny access completely. This opt-in design means a standard Windows 11 installation remains unchanged unless the user deliberately enables agentic features and authorizes specific assistants.

Each artificial intelligence assistant receives its own configuration area within Windows settings, giving users a way to revisit and adjust permissions after initial setup. Alongside these folder protections, Microsoft is also testing discrete connectors that govern how agents interact with system applications such as File Explorer and Settings, which are managed separately from core personal folder access rules. This modular approach is intended to let users allow an artificial intelligence agent to carry out actions like changing system configurations while still preventing that same agent from viewing or manipulating personal content such as photos or documents.

53

Impact Score

PaTH Attention boosts large language model positional reasoning

Researchers at MIT and the MIT-IBM Watson Artificial Intelligence Lab have introduced PaTH Attention, a new positional encoding method that makes transformers more context-aware and better at tracking state over long sequences. The technique adapts position information based on token content and can be combined with forgetting mechanisms to improve long-context reasoning and efficiency.

China reportedly tests domestically built euv lithography prototype

China has reportedly built and begun testing a domestically developed euv lithography prototype assembled from second-hand components and reverse-engineered designs. Huawei is leading a broader effort to create a fully domestic artificial intelligence semiconductor supply chain spanning chip design to advanced manufacturing tools.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.