Meta launches Muse, a personal AI agent built to act across the web
Meta is rolling out Muse in the United States, pairing a task-performing agent with a dedicated cloud computer, approval checkpoints and user-controlled connections.
Meta has begun rolling out Muse, a personal AI agent designed to do multi-step work rather than simply return answers. Available in the United States through iOS, Android, muse.ai and WhatsApp, Muse can browse sites, complete forms and keep working after a user closes the app.
The agent runs on a dedicated Muse Secure VM and is powered by Meta’s Muse Spark model. Users choose which services it can connect to and whether it may only read data or also take actions. Muse asks for approval before sensitive steps such as sending email or making a purchase, and provides an audit trail.
The launch makes Meta a direct participant in the market for persistent, action-taking consumer agents. It also raises the stakes for the company’s privacy architecture because Muse can remember personal context and operate across accounts. Meta says interactions are excluded from ad systems and users may opt out of model training.
Source: Meta Newsroom ↗