Meta is asking users to hand it far broader access to their daily digital life by launching Muse, a personal AI agent that connects to email, calendars, payments, health and fitness apps, smart home systems, shopping and entertainment services. That shift changes what consumers must weigh: convenience from an assistant that can act on their behalf, against broader exposure of highly sensitive personal data to Meta.
Muse is designed to perform tasks for users, the company says, including sending emails, booking travel, lowering bills, filling out forms, creating plans, converting recipe reels into grocery lists, sending invites and completing purchases using Link by Stripe, which offers purchase protections. Shopify’s Shop Pay and 1Password integrations are slated to arrive soon. Meta says people pick which apps to connect, one at a time, and Muse can connect via public APIs or by accessing services through a browser when no API is available.
The agent runs on Meta’s Muse Spark model and will be available on the web at muse.ai, Apple and Android apps, and in WhatsApp chats, with plans to extend to the company’s AI glasses. Meta will offer a free tier, while higher use will be metered and can be covered by paid subscriptions, Power at $20/month or Maximum at $100/month. A payment card is required to get started, the company says, and the app includes a usage meter and warnings when free allowances run out.
To address privacy and security concerns, Meta describes Muse operating inside a dedicated secure virtual machine it calls Muse Secure VM, with a separate Sentinel agent kept isolated at the system level. The company asserts Muse will not see passwords or payment credentials and that conversations and user data are not fed into Meta’s advertising systems. These are engineering claims that, Meta acknowledges, will need independent scrutiny by security experts.
Those technical assurances arrive as Meta faces heightened skepticism. Less than two weeks before the Muse launch the company agreed to a multistate settlement valued at $18 billion in litigation over consumer harms. The company also carries a history of regulatory actions and incidents, including a $5 billion FTC penalty in 2019, a 2011 FTC settlement over privacy practices, a 2023 FTC charge for violating a privacy order and past exposures of user passwords and third-party misuse of platform data. That record shapes how many consumers and regulators will view Muse’s request for deeper access.
The immediate questions are adoption and oversight. Meta must show that its VM architecture and operational controls keep sensitive credentials and conversations out of risky pipelines, and regulators and security researchers will likely probe those claims. If users adopt Muse widely, Meta will gain more direct signals about everyday behaviour and transactional intent; if trust does not follow, the product may struggle to scale beyond early adopters.
