Meta Launches Muse Personal AI Agent App With $20-$100 Subscription Tiers Amid Privacy Scrutiny

Meta introduced its Muse personal AI agent app on Tuesday, offering a free tier alongside paid subscription plans priced at $20 or $100 per month depending on usage. The app, internally code-named Hatch, is powered by Meta's Muse Spark family of foundation models and is designed to handle tasks such as booking appointments, filling out electronic forms, and monitoring home security camera feeds.

Meta AI chief Alexandr Wang told CNBC the app is built to feel "approachable and friendly and explainable" while performing complex work behind the scenes. "Behind the scenes, Muse might be doing very advanced coding workflows, or building sophisticated integrations, or doing quite a lot of heavy lifting while keeping that very sort of simple for the user," Wang said.

Availability and access

The Muse personal agent will be available to U.S. consumers via iOS, Android, and a standalone website, with plans to eventually support Meta's Ray-Ban Meta glasses. A version will also run through WhatsApp, though that version omits app-only features like a personalized feed and an "ideas" tool that recommends AI agent activities. According to Wang, the core conversational functionality remains consistent across both the app and WhatsApp.

Privacy and security claims

Wang said the app operates within "its own isolated environment" inside Meta's computing infrastructure and "never sees your actual passwords or payment details," asking users before taking sensitive actions. These are Meta's claims and have not been independently verified. Meta says it has hardened the product through internal testing, "agentic red teaming," and a private bug-bounty program that pays security researchers for identifying vulnerabilities, according to a company technical blog post.

Meta VP of Engineering David Singleton said users must opt out if they don't want their Muse interactions used to train the company's AI models; otherwise, Meta will scrub "critical personally identifying information" before using conversations for model improvement.

Monetization plans

Wang said Meta is exploring taking a cut of AI agent-driven shopping transactions but has not finalized a business model. "We think the commerce business model is potentially really interesting for this product," Wang said.

Context: scrutiny and industry pressure

The launch lands during what CNBC describes as a fraught period for Meta. The company recently agreed to pay nearly $17 billion to settle claims from a coalition of state attorneys general alleging it misrepresented harms on Facebook and Instagram, and it still faces personal injury and school district lawsuits on similar grounds. Separately, the AI industry broadly faces rising concern over cybersecurity risks tied to autonomous agents, alongside public backlash against AI data center buildouts.

Zuckerberg has told investors that personal agents represent Meta's next major AI bet, saying on the company's July earnings call that such tools would form "the foundation for our next wave of products and revenue lines." Wang acknowledged Meta is "pretty early in this new era of personal agents" despite positioning Muse as more broadly accessible than rival offerings from OpenAI, Google, and startups.

What this means

Meta is betting that consumer-facing agentic AI can become a new revenue line as it seeks to justify massive infrastructure spending and diversify beyond advertising. The $20-$100 pricing tiers suggest Meta expects heavier agentic workloads to carry meaningfully higher compute costs. However, launching an autonomous agent with access to passwords, payments, and security cameras—precisely as Meta absorbs a $17 billion privacy settlement—invites immediate scrutiny of whether its isolation and opt-out safeguards hold up against real-world adversarial testing, not just internal red-teaming.