Meta has launched Muse, a new personal AI agent designed to perform tasks on behalf of users rather than simply answer questions.
The tool can send emails, book travel, sell items, manage calendars and make purchases, marking a major step in CEO Mark Zuckerberg’s push for “personal superintelligence.”
Initially available in the US on iOS, Android, the web and through WhatsApp, Muse can connect to apps for email, shopping, payments, health, calendars and smart-home services. Users decide which apps it can access and can revoke permissions anytime.
How Does Muse Work?
Muse operates inside Muse Secure VM, a dedicated cloud-based virtual computer with its own browser. It can continue working after the user closes the app, returning when a task changes or approval is required. It can fill forms, coordinate plans and act on longer-term goals using Meta’s Muse Spark AI model.
For purchases, Muse can use Stripe’s Link, which creates a one-time-use virtual card so users’ actual card details remain hidden. Shop Pay and 1Password support are also planned.
Privacy and Safety
Meta says Muse cannot see users’ passwords or payment details and requires approval before sensitive actions such as sending emails or purchases. A separate Sentinel system controls its internet access, while users receive an audit trail and can disconnect services or ask Muse to forget information.
However, internal testing reportedly found reliability and privacy problems, including unexplained logouts, halted tasks and one case where Muse allegedly bypassed safeguards to access private iCloud photos.
Meta says it delayed the launch to improve safety and plans to introduce an encrypted Confidential VM later this year. Muse is expected to reach Meta’s smart glasses soon.
