Microsoft has introduced a new feature in its artificial intelligence system that could pose a serious risk to online privacy. Following the launch of Recall, the tech giant has now unveiled Copilot Vision, a tool designed to understand what you are viewing on a website and interact with that information. This feature is currently available to select users through the Copilot Pro subscription.
Please follow us on Facebook and Twitter.
Announced in early October as part of a major update to Microsoft’s AI, Copilot Vision is integrated into the Edge browser. It allows users to interact with AI while browsing the web. The tool can analyze what appears on the browser screen and respond to specific questions using voice commands.
Unlike simply opening the sidebar to summon Copilot and typing a question, Vision provides a hands-free experience. Users can use their voice to ask questions about concepts they don’t understand on the web or seek advice from Copilot, such as how to plan a trip or select the best home decor.
To deliver this functionality, Microsoft’s AI leverages Copilot Voice, a feature similar to ChatGPT’s voice assistant that enables smooth, conversational interactions. When paired with Copilot Vision, the browsing experience becomes more immersive and interactive. However, there is one small concern: privacy.
Microsoft AI Analyzes and Understands What You See on Your Screen
The key feature of Copilot Vision is also its most debated aspect. Its ability to analyze and understand everything you view in the browser means the AI has access to your screen content. However, Microsoft claims that interactions with Vision and Voice are temporary and are not saved on its servers.
“We’ve prioritized security, control, and privacy above all else, ensuring that every time you use Vision, your data is handled and protected in accordance with the Microsoft Privacy Statement,” the company states in the Copilot Vision FAQ section.
While Microsoft does not store interactions with Copilot Vision, the model’s responses are recorded to monitor and prevent unsafe interactions and outcomes. The company assures users that it does not capture end-user input, images, or browsing context.
Another important point is that Copilot Vision is not always active, allowing users to browse without concern. The tool is only enabled at the user’s request on any website. Users can activate the Vision icon whenever they want to start a session, which ends when they close the Edge page or click the “X” button in the interface.
Similar to Recall, it is entirely up to the user to decide whether to allow the AI to analyze everything on their screen. Currently, Copilot Vision is available to a limited group of users in the United States and requires a Copilot Pro subscription. Microsoft stated that this phase will be used to test features and collect feedback for a potential wider release.