Particle.news

Meta Launches Muse, a Personal AI Agent That Acts for Users

Meta says Muse will let people hand everyday tasks to an AI using Secure VM and Sentinel controls with a user‑held Confidential VM promised later this year.

Overview

  • Muse launched in the U.S. on Tuesday as an app, on the web at muse.ai, and inside WhatsApp, and Meta plans support for its AI glasses soon.
  • The agent can autonomously perform multi‑step chores such as sending emails, booking travel, filling forms, lowering bills, creating shopping lists, and making purchases using Stripe’s one‑time Link cards.
  • Each Muse runs inside a dedicated cloud workspace called Muse Secure VM with a separate Sentinel monitor that vets internet outputs and prompts users for approval before sensitive actions.
  • Meta offers a free tier plus paid plans called Power ($20/month) and Maximum ($100/month) to raise usage limits, and says Shop Pay and 1Password integrations will arrive soon.
  • The rollout tests new privacy engineering and a business model but faces a trust hurdle because Meta recently settled multistate suits and critics will watch whether the promised Confidential VM, audits, and real‑world safety checks materialize.