Meta launches Muse AI assistant for U.S. consumers

Meta unveiled Muse, a personal AI assistant for U.S. users that can interact with email, calendars, payment services, health apps, smart-home devices and other everyday applications. The agent, built on Meta’s Muse Spark model, can draft messages, book travel, generate grocery lists, make purchases through a Stripe-based checkout and perform other routine tasks.

Users must link individual apps or services one at a time, giving them control over which data the assistant can access. Connections are made via built-in integrations, public APIs or, when no API exists, through a browser interface. Muse is accessible through a web portal, iOS and Android apps, and WhatsApp chats, with plans to add support for Meta’s AI glasses.

The service is free at launch, but Meta will offer two subscription tiers—Power at $20 per month and Maximum at $100 per month—as usage exceeds the free allowance. A usage meter will alert users when they approach their limit and present upgrade options.

Meta says Muse runs in a dedicated, secure virtual machine that isolates the assistant from passwords and payment details, and that conversation data are not shared with the company’s advertising systems. A separate Sentinel agent operates on the same machine but remains isolated at the system level.

The rollout follows Meta’s recent $18 billion settlement with 29 states over alleged consumer harms. Critics note the company’s past privacy violations, including FTC penalties in 2011, 2019 and 2023, and the 2018 Cambridge Analytica scandal, raising questions about whether users will grant the extensive data access required by Muse.

To encourage adoption, Meta allows users to personalize the assistant with a name, avatar and communication preferences, aiming to build a sense of connection while delivering functional benefits.