ThynkRise Meta Muse is a personal AI agent that runs errands for you. You give it a goal, and it sends the emails, books the travel, fills out the forms and pays at checkout. Meta announced Muse on September 8, 2026, and it’s available now in the US. For most people the harder decision is whether to give an AI agent from Meta access to their email, calendar and payment details. Below: what Muse can do, what it costs, how Meta says it protects you, and who should try it now or wait.
What Meta Muse can do
Muse runs on Muse Spark, which Meta calls its most capable model for “real-world agentic work.” The difference from a chatbot is that Muse acts. You describe a goal, and it works through the steps across the apps and accounts you connect.
At launch, Meta says Muse can draft and send emails, book travel and appointments, and open a browser to fill out forms for you. It can turn a long-term goal into a step-by-step plan and keep working on it after you close the app. It also shops. Meta’s example is turning a recipe reel you saved on Instagram into a grocery list, and Muse can suggest purchases and handle the checkout. Meta says Muse can negotiate on your behalf too, but it hasn’t said much about how that works.
Checkout runs through Link, Stripe’s payment wallet, which Meta says adds coverage for damaged or lost items, price drops and no-fee returns. Meta says Shop Pay and 1Password support are coming later.
What Muse costs and where you can use it
Muse launched in the US on the web at muse.ai, in iOS and Android apps, and inside WhatsApp chats. Meta says support for its AI glasses is coming. According to TechCrunch’s launch coverage, there are three tiers:
- Free, with usage limits (Meta says this covers most of what people need)
- Power, $20 a month
- Maximum, $100 a month
The paid tiers are priced by how much you use Muse. You have to add a payment card to get started, even on the free tier.
How Meta says it keeps your data and money safe
Muse can spend money and send messages in your name, so its safeguards matter more than its feature list. According to Meta:
- Muse asks before sensitive actions, such as sending an email or making a purchase.
- You can see an audit trail of everything it has done and everything it plans to do next.
- Each person’s agent and data live on their own “Muse Secure VM,” a dedicated virtual machine with its own browser.
- A separate “Sentinel” agent has to approve every action that goes out to the internet.
- Muse can’t see your passwords or payment methods directly.
- A “Muse Confidential VM,” encrypted with keys only you hold, has been announced but isn’t available yet.
On privacy, Meta says you can opt out of your conversations being used to train its AI, that Muse doesn’t share your conversations or the data in your VM with Meta’s ad systems, and that you can disconnect any service or tell Muse to forget something at any time.
The trust question
These are Meta’s own descriptions. Independent security researchers haven’t verified them yet, and TechCrunch’s launch coverage says the claims need exactly that kind of scrutiny. The context is hard to ignore: Meta paid a record $5 billion FTC penalty over privacy violations in 2019, was charged again by the FTC in 2023 with violating that privacy order, and settled an $18 billion multistate lawsuit over harms to minors in August 2026, weeks before Muse launched.
That history doesn’t prove Muse is unsafe, but it does put the burden of proof on Meta, and it makes early users the test group. CNBC framed the launch as a public test of whether people will trust personal AI agents with this level of access at all.
Who should try Muse now, and who should wait
Muse is a reasonable experiment if you spend real time on repeat digital errands like scheduling, forms and reorders, you don’t mind reviewing its plan before each sensitive step, and you’re willing to start with low-stakes accounts.
Wait if getting value from it would mean connecting your main email, your primary bank card or your health apps, or if you want independent security reviews first. Waiting a few months costs you very little, and by then outside researchers will have had time to check Meta’s claims.
If you’re weighing AI assistants more broadly, our guides to GPT-6 Astra and Gemini 3.1 Pro cover the chat-first alternatives, and the AI Tools & Automation section has the rest.
How to try Muse carefully
- Start on the free tier, and only pay for Power or Maximum once you know which chores you’ll hand over.
- Connect as little as possible at first, such as your calendar or a secondary email account.
- Check the audit trail before you approve anything, especially purchases.
- Turn off AI training in the privacy settings if you don’t want your conversations used to train Meta’s models.
- Once a month, look at which services are connected and disconnect anything you no longer use.
Frequently asked questions
Is Meta Muse free?
There’s a free tier with usage limits. Heavier use costs $20 a month (Power) or $100 a month (Maximum), and you need a payment card on file to start.
Is Meta Muse available outside the US?
Not yet. It launched in the US only, on the web, iOS, Android and WhatsApp.
Does Muse use my data for Meta’s ads?
Meta says Muse doesn’t share your conversations or the data in your Muse VM with its ad systems, and that you can opt out of your chats being used to train its AI.
Is this the same as Microsoft’s Muse?
No. Microsoft Research also has an AI model called Muse, built to generate video game environments. Meta Muse is a separate, consumer-facing personal agent, and the two are unrelated.
