I tried Meta’s new Muse AI agent — it’s incredibly useful, but handing it my digital life felt deeply uncomfortable

Meta's new Muse AI platform wants access to your digital life to help manage it for you. The AI agent can complete tasks like sending emails, filling out forms, shopping, and more. Muse is available as an app in the U.S., through WhatsApp, or via the Muse.ai portal. It works by opening a browser in its own virtual computer and moving between connected services, returning when it has finished or needs permission to perform sensitive actions.

The experience of using Muse feels both convenient and uncomfortable. The AI agent asks for permission to sign into your online accounts, which can feel intrusive before it begins. Users can try Muse for free, with the option to pay $20 per month for the Power plan or $100 for the Maximum plan. After choosing a nickname, Muse asks what you'd like it to do for you. For example, it can help with shopping by suggesting options based on your size and finding deals on products.

When Muse handles sensitive tasks like email, users must grant it permission to access their inbox, write on their behalf, and learn more about them through their emails. While this can streamline online errands, it also feels more personal. Meta has implemented security measures, such as filtering out one-time tokens and password-reset links, to protect users' information.

The Confidential VM mode is in development, which would cryptographically prevent Meta from accessing information within the virtual machine. Despite the convenience Muse offers, the more it can assist, the more personal information it requires, making users cautious about granting access.