Meta Debuts Muse AI Agent, but Privacy Record Threatens Consumer Trust
Meta launches Muse, a personal AI agent for email, payments and purchases, but its broad data access and privacy record could test U.S. consumer trust.
Summary
Meta introduced Muse in the U.S. on September 8, 2026, less than two weeks after its $18 billion August settlement with 29 states over social media harms. Powered by Muse Spark, the opt-in agent connects individually to email, calendars, payments, health, fitness, smart-home, dining, shopping, music and event services. It can send email, book travel, lower bills, fill forms, create plans, convert recipe reels into grocery lists, send invitations and buy goods. Checkout uses Stripe’s Link with purchase protection; Shopify Shop Pay and 1Password integrations are coming. Unsupported services can connect through public APIs using supplied credentials or through a browser when no API exists.
Muse is available through muse.ai, iOS, Android and WhatsApp, with Meta AI glasses support coming. Free access requires a payment card; Power costs $20 monthly and Maximum costs $100, while a meter and alerts track usage. Meta expects most users to remain free. Muse continues tasks after users leave, learns priorities from conversations, makes unprompted suggestions and supports customized names, avatars, appearances and communication settings. Competing approaches include Gemini Spark, Claude Cowork and agents integrated with iMessage, SMS and WhatsApp.
Meta says each Muse runs in a dedicated Muse Secure VM alongside a system-separated Sentinel agent, cannot see passwords or payment methods, and keeps conversations and data from Meta’s advertising systems. Those claims await independent security scrutiny. Trust remains the obstacle: Instinct testers encountered a sweeping perpetual, irrevocable content license; Meta settled FTC deception charges in 2011, paid a record $5 billion for eight privacy violations in 2019, faced another FTC charge in 2023, discovered readable passwords in 2019 and remains shadowed by Cambridge Analytica. Beyond the 29-state deal, a New Mexico child-harms case cost $942 million, while thousands of personal-injury and school-district cases remain pending.
Positives
- Muse can complete practical tasks including booking travel, lowering bills, filling forms, planning events and making protected purchases through Stripe’s Link.
- Users connect apps and services individually, making access explicitly opt-in rather than granting Muse everything at once.
- Muse Secure VM separates the Sentinel security agent and prevents Muse from seeing passwords or payment methods, Meta says.
- Free access spans the web, iOS, Android and WhatsApp, with Meta AI glasses support planned.
- Public API and browser options allow Muse to work with services lacking built-in connectors.
Risks & concerns
- Muse may access highly sensitive email, calendar, payment, health, fitness and smart-home information.
- Meta’s record includes 2011 and 2023 FTC actions, a $5 billion privacy penalty, readable passwords and Cambridge Analytica.
- Independent security experts have not yet fully tested Meta’s claims about Muse Secure VM and advertising-system separation.
- Even free-tier users must provide a payment card before using Muse.
- Meta recently agreed to an $18 billion settlement with 29 states and paid $942 million in a New Mexico child-harms case.
- Muse learns from conversations and can make unprompted suggestions, increasing the importance of dependable privacy controls.