Meta's new AI agent Muse faces an uphill battle earning consumer trust

Meta has entered the agentic AI race with Muse, a personal assistant designed to handle everyday tasks directly rather than simply answer questions. Available starting today via the web, iOS, Android, and WhatsApp, Muse represents the company's most ambitious consumer AI bet yet—though it arrives at a fraught moment for the social media giant.

What Muse does#

Muse connects to your everyday digital life: email, calendars, payments, health apps, smart home systems, and more. The AI can send emails, book travel, negotiate lower bills, fill out forms, create itineraries, convert recipe videos into shopping lists, and make purchases through Link by Stripe's secure checkout. The service is free to start, with paid tiers—Power ($20/month) and Maximum ($100/month)—kicking in as usage climbs.

The agent operates continuously in the background, learning from your preferences to offer suggestions without prompting. It's available across devices and will eventually ship on Meta's AI glasses. Users maintain granular control, connecting each service individually rather than granting blanket access.

The trust problem#

Meta's announcement comes just two weeks after settling an $18 billion multistate lawsuit over social media's documented harms to children—a context that casts a long shadow over Muse's privacy and security claims.

The company's track record complicates adoption efforts: the FTC settled with Meta in 2011 for deceiving users about privacy protections, imposed a record-breaking $5 billion penalty in 2019 for eight separate privacy violations, and charged Meta again in 2023 for violating a prior privacy order. The 2019 discovery that Meta stored user passwords in readable text, and the lingering Cambridge Analytica scandal in which millions' data was harvested without consent, further erode confidence.

Despite these precedents, Meta insists Muse operates differently. The agent runs in its own dedicated, isolated virtual machine that blocks access to passwords and payment details. A separate Sentinel agent handles verification. Meta also claims Muse conversations and data remain walled off from its advertising systems.

Building connection to overcome skepticism#

To counter doubts, Meta has published detailed technical documentation explaining its security approach and has designed Muse to feel more personal—users can name their agent, customize its avatar, and configure how it communicates.