What changed: Meta entered the personal AI agent market directly with Muse, extending its reach well beyond social media.
Meta has announced Muse, a personal AI agent powered by the company's in-house Muse Spark model. Designed to handle everyday tasks, Muse can send emails, book travel, manage bills, and make purchases on a user's behalf. Given a goal, Meta says the agent can work autonomously — opening a browser, filling out forms, and even negotiating on the user's behalf.
Muse connects to email and calendars as well as a wide range of third-party services, including health apps, smart home platforms, dining, shopping, and event apps. For lengthy tasks, the agent continues working in the background after the app is closed, returning to the user when something changes or when approval is needed — such as completing a purchase. Checkout is handled through Stripe's Link, with 1Password and Shop Pay integrations planned.
Muse is now available in the US on iOS, Android, and muse.ai, with WhatsApp access also available. Support for Meta's AI glasses is coming soon. Getting started requires a payment card; the app includes a usage meter showing what percentage of the free allowance remains and presents subscription options when that allowance runs out. Two paid plans — Power at $20 per month and Maximum at $100 per month — target heavier users, though Meta expects most people to remain on the free tier.
For security, Meta designed a dedicated cloud environment called Muse Secure VM. The company says the agent has no visibility into users' passwords or payment methods. A separate agent called Sentinel runs on the same virtual machine and is tasked with blocking any unauthorized internet access by Muse. Meta also announced plans for a confidential VM variant later in 2026, encrypted so that even Meta cannot access it.
On privacy, Meta claims Muse does not share user conversations or data with its advertising systems. Users can opt out of having their interactions used to train Meta's AI models, though whether this option is on or off by default remains unclear. Users can also instruct Muse to forget specific things it has learned about them.
The timing of the announcement is notable: Meta introduced Muse less than two weeks after agreeing to an $18 billion multistate settlement over social media consumer harms. The company's history — a 2011 FTC settlement over deceptive practices, a then-record $5 billion penalty in 2019, and a new FTC charge in 2023 — places its privacy assurances for Muse under heavy scrutiny. The Muse Secure VM architecture has not yet been independently verified by security researchers.
Key facts
- Muse runs on Meta's in-house Muse Spark model and can handle everyday tasks such as sending emails, booking travel, managing bills, and shopping.
- Available in the US on iOS, Android, muse.ai, and WhatsApp; AI glasses support is coming soon.
- A free tier is available alongside two paid plans: Power and Maximum.
- Muse Secure VM and a companion Sentinel agent form the security layer; Muse has no access to passwords or payment methods.
- The announcement came less than two weeks after Meta agreed to an $18 billion multistate settlement over social media consumer harms.
Why it matters
Muse requires access to sensitive data including email, payments, and smart home systems, raising significant trust questions given Meta's long history of privacy violations.
Watch next: Independent security audits of Muse Secure VM and the timeline for expansion beyond the US are the key developments to track.
Source date: 2026-09-08T21:21:02.509Z · Verification confidence: 85/100
