MuseWork
Get Started for Free
Back to Blog
AI NewsMeta MuseAI agent

What Is Meta Muse? Features, Pricing, Privacy, and Availability

What Is Meta Muse? Features, Pricing, Privacy, and Availability

Meta Muse is one of the clearest signs yet that AI assistants are moving beyond chat. Announced on September 8, 2026, Muse is Meta’s personal AI agent: you give it a goal, and it can use a browser and connected services to move the task forward.

The idea is compelling. So are the questions it raises. What can Meta Muse actually do? Is it free? Where is the app available? And how much access do you have to give it before it becomes useful?

Here is the quick answer:

  • Meta says Muse can browse websites, work with connected services, and keep some tasks moving after the app closes.
  • Under Meta’s described defaults, it stops for actions that are difficult to reverse, such as sending an email or completing a purchase; permission settings can vary.
  • Meta says Muse is free for most needs and offers subscriptions for more usage, but it has not published a complete public pricing table.
  • The current Muse Secure VM separates credentials and permissions from the main agent, but it does not technically prevent Meta from accessing data when needed to operate, secure, or support the service.
  • Interaction “trajectories” may be used for model training by default. Users can opt out in Muse settings.

Most available evidence still comes from Meta’s own launch and technical materials, not long-term independent testing. This guide separates the product pitch from the details that matter before you connect an account or hand over a real task.

Disclosure: MuseWork and Meta Muse are built by different companies and different teams. MuseWork is not affiliated with, endorsed by, or sponsored by Meta or Meta Muse.

What is Meta Muse?

Meta Muse is a consumer-focused personal AI agent powered by Meta’s Muse Spark model family. You can talk to it through the Muse app or WhatsApp. Instead of stopping at an answer, Muse is designed to plan steps, use tools, and take action on your behalf.

Each Muse runs in a dedicated cloud environment called Muse Secure VM, with its own browser, file system, storage, and task tools. Meta says credentials and permission decisions are separated from the main agent runtime.

One distinction is worth keeping in mind: Muse is the consumer agent, while Muse Spark is the model family behind it. Muse Spark benchmarks and API pricing are not consumer subscription details for the Muse app.

What can Meta Muse do?

The simplest way to understand Meta Muse is as a bridge between a conversation and the work that follows it.

Work through websites

Meta says Muse can search the web, navigate sites, fill in forms, and complete transactions. Launch examples include booking travel, shopping, selling a car, negotiating a bill, and sending email. Users can watch its browser and take over when needed.

Those examples show what Meta intends Muse to do. They do not yet prove that every unfamiliar website or multi-step workflow will work reliably.

Use connected services—with your permission

Meta’s official App Store and Google Play descriptions list email, calendar, Instagram, Facebook, health and fitness data, and finances. Users choose what to connect and what access to grant.

That list should not be read as a guarantee that every connection is available to every account. Support may vary by service, platform, region, and account, so check the options shown inside the product.

Stay with longer-running tasks

Meta says Muse can manage multiple tasks, continue some work after the app closes, and notify users when something changes or input is needed. A Goals view helps track longer-running objectives, while editable memory files and an activity trail show what Muse remembers, has done, and plans to do.

This is where the product becomes more than a browser demo. Continuity can save time—but only if the context stays accurate. For important work, review the memory and activity trail rather than assuming every stored detail is current.

Stop before hard-to-reverse actions

Under Meta’s described defaults, Muse stops before actions such as sending an email or making a purchase. Its separate Sentinel system evaluates connector actions and outbound requests, then allows, denies, or asks the user according to the applicable permissions. Meta also says some permission settings can be made more or less cautious.

At launch, Muse can check out through Link by Stripe. Meta says Link creates a one-time-use card so the merchant does not receive the user’s regular card number. Shop Pay and 1Password support are planned for later and should not be treated as generally available launch features.

Build a tool when the task needs one

Meta’s product materials say Muse can write code and build tools a task needs. The public materials do not define the review process, permissions, or limits of this capability, so it is better understood as an announced capability than a fully documented development platform.

Meta Muse pricing: what is official, and what is reported?

Meta says Muse is free for most of what people need, with subscription plans for people who want more usage.

What is missing is a complete public Meta pricing table. The official pages reviewed for this update do not specify included usage, billing terms, or cancellation conditions.

TechCrunch reported two launch plans: Power at $20 per month and Maximum at $100 per month. It also reported that a payment card is required to get started. Those numbers are launch-day media reporting, not prices verified on a public Meta pricing page. Check the current amount and terms inside the official product before subscribing.

Is there a Meta Muse app, and where is it available?

Meta offers Muse through iOS, Android, the web at muse.ai, and WhatsApp. Both official app-store listings identify the developer as Meta Platforms, Inc. Check the developer before downloading, since many unrelated apps also use “Muse” in their name. Official source links are included at the end of this article.

As of September 9, 2026, Muse is rolling out in the United States. A rollout does not mean every account or device receives access immediately, and Meta has not published a country-by-country expansion schedule.

The U.S. App Store lists the app as 18+, while Google Play gives it a Teen content rating. These labels come from different store-rating systems and are not a universal Meta account-age rule. Check the official sign-up flow and applicable terms for current eligibility requirements.

Meta also says support for its AI glasses is coming, but has not announced a firm public date.

Meta Muse privacy: the safeguards and the tradeoffs

An agent that can browse, remember, connect to accounts, and make transactions needs more than a privacy promise. Meta has published details about Muse’s safeguards—and some important limits.

The main agent is designed not to see real credentials

In Muse Secure VM, credentials and authentication tokens are stored in an isolated component. Meta says the main agent receives substitute credentials for authorized actions instead of seeing the underlying secrets. Sentinel separately enforces connector and network permissions.

That reduces exposure; it does not remove risk. Meta says prompt injection remains an open problem and that Muse “can and will still make mistakes.”

Meta can still access current Secure VM data when necessary

The launch version of Secure VM does not technically make user data inaccessible to Meta. Meta says staff access is restricted by operational policy, but the architecture does not prevent access when needed to support, secure, or operate the service.

Meta plans a stronger Muse Confidential VM later in 2026. It is intended to use a user-held key to prevent Meta from accessing VM contents. That is a future capability, not a protection generally available at launch.

Model training is opt-out

Meta says conversations, tool calls, and related interaction “trajectories” may be sanitized to remove key personally identifiable information and used to train future models by default. Users can opt out in Muse settings.

Meta also says conversations and VM data are not shared with its ad systems. But when Muse visits a retailer or uses another service on a user’s behalf, that activity can be treated as the user’s own and may still influence ads indirectly.

Before connecting a sensitive personal or work account, check both the training setting and the permissions for that service.

Meta Muse vs. MuseWork: which one fits the work you need done?

The names are similar. The focus is not.

Meta positions Muse as a personal agent for tasks across everyday life. MuseWork is built for professional work: the same AI work partner stays with your goal and context, carries multi-step work forward, produces results you can review, and continues from your feedback instead of asking you to rebuild the brief every time.

A practical way to choose:

  • Explore Meta Muse if you want to evaluate a personal agent that works across consumer services and everyday tasks.
  • Choose MuseWork if your priority is sustained professional collaboration—from an objective to a reviewable result, then into the next round of work.

MuseWork and Meta Muse are built by different companies and different teams. There is no corporate affiliation, shared product team, endorsement, or sponsorship between them. The comparison below is about product focus and use cases. For a closer look, read the full Meta Muse vs. MuseWork comparison.

Frequently asked questions

Is Meta Muse free?

Meta says Muse is free for most needs and offers subscriptions for more usage. TechCrunch reported $20 and $100 monthly launch plans, but Meta has not published a complete public pricing table. Verify current pricing inside the official product.

Is Meta Muse available outside the United States?

Not on a confirmed general basis. Meta says Muse is rolling out in the United States and has not published a country-by-country expansion schedule.

Can Meta Muse send emails or make purchases?

Meta says it can when the necessary service and permissions are available. Under the defaults Meta describes, Muse stops for hard-to-reverse actions such as sending an email or making a purchase; exact handling can depend on permissions.

Can Meta Muse see passwords or payment details?

Meta says the main agent does not see real credentials. They are stored separately and inserted for authorized actions. This is an architectural safeguard, not a guarantee against every security failure.

Does Meta use Muse activity to train AI models?

Meta says sanitized interaction trajectories may be used for model training by default. Users can opt out in Muse settings.

Is Meta Muse the same as Muse Spark?

No. Muse is the personal AI agent; Muse Spark is the model family that powers it.

Is Meta Muse connected to MuseWork?

No. MuseWork and Meta Muse are products from different companies and different teams. MuseWork has no corporate affiliation with Meta and is not endorsed or sponsored by Meta.

The takeaway

Meta Muse matters because it pushes the assistant experience from “tell me how” toward “help me get it done.” Its browser, memory, background work, and connected services make that direction tangible. The real test will be whether it can deliver that convenience consistently while keeping permissions and approvals understandable.

For now, evaluate it with limited access, a reversible task, and a close look at its settings. Pricing, connector coverage, international availability, and independent evidence of day-to-day reliability are still limited.

If the problem you are solving is specifically professional—keeping context across a multi-step task, reaching a result you can inspect, and continuing from feedback—review the full Meta Muse vs. MuseWork comparison, then start working with MuseWork when you are ready.

Sources