Overview
- Muse launched as a personal AI agent on Sept. 8 and has quickly reached millions of installs, topping U.S. app charts and helping lift Meta’s stock by more than 20 percent in the weeks after release.
- Meta opened a connector portal that lets third‑party developers submit integrations to Muse, a move that produced early commercial ties such as Shopify enabling Shop Pay while Amazon has blocked Muse from parts of its shopping site.
- Meta runs each user agent inside a per‑user Muse Secure VM and uses a Sentinel supervisor and one‑time payment cards to limit exposure, and the company says it will add a Confidential VM that stores data encrypted under a user‑held key later this year.
- Security scrutiny intensified after a reported macOS vulnerability showed local apps could alter undocumented Muse settings and, according to researcher Patrick Wardle, could be used to capture authentication tokens and manipulate agents without obvious user warnings.
- Analysts see Muse as a potential new revenue engine through subscriptions and transaction share, but experts warn merchant blocks, unresolved developer commercial terms, ongoing security flaws, and privacy concerns could constrain how widely and quickly Muse becomes a platform.