In early October, Microsoft announced Copilot Vision, a new feature that could harness the power of AI to help you complete various tasks by analyzing the content on the screen and providing the necessary assistance. Two months after the initial announcement, Microsoft is finally launching Copilot Vision in preview.
When enabled, Copilot Vision "sees" and understands what is happening on the screen and its context. You can ask Copilot questions about the web page, and the assistant will answer and provide additional information.
Copilot Vision is currently available in a limited preview to Copilot Pro subscribers in the United States. Also, the feature only works in the Microsoft Edge browser and only on certain websites. Microsoft says it wants to ensure Vision is a "useful and safe experience" from the get-go, but the company plans to expand the number of available websites as the testing scope broadens.
Microsoft also stresses that Copilot Vision does not collect or store any user data, nor does it use it for model training. Once the session is over, Copilot Vision will delete all the data about what you said or what information you shared. Like the recently launched Recall (also in preview for Windows Insiders with Copilot+ PCs), Copilot Vision is an opt-in feature—it will not invoke unless you launch it.
Additionally, Microsoft promises to prioritize copyright, privacy, and safety. It will not offer any information on paywalled websites, and it will not be able to take action on your behalf. The sole purpose of Copilot Vision is to "have a second set of eyes" and a helpful assistant who can tell you more about whatever you ask.
You can access Copilot Vision Copilot Labs. Getting it to work requires a Copilot Pro subscription in the United States. You can learn more about Copilot Vision in a dedicated FAQ section on the official website.