Meta’s new AI can read your emails, spend your money and work while you sleep
Muse can send emails, book trips and make purchases — but Meta still has to win your trust
Imagine telling an AI to organise your holiday, then closing the app.
While you get on with your day, it checks flights, builds the itinerary, and keeps working in the background. It could later fill out forms, send an email or even make a purchase — stopping for your approval when needed.
That is the idea behind Muse, Meta's new personal AI agent, designed to move artificial intelligence beyond answering questions and into actually doing things for people.
Meta launched Muse in the US on Tuesday, pitching it as an AI that can handle everyday tasks and work towards longer-term goals.
But to become genuinely useful, Muse needs something more valuable than a clever prompt: access to your digital life.
More than another chatbot
Muse represents the industry's shift from AI assistants that respond to prompts towards "agents" capable of taking actions independently.
Users can ask Muse to send emails, book travel, fill out forms, shop online or work towards longer-term goals.
It has its own browser and virtual computer and can continue working after the user closes the app. Meta says Muse can use a file system and terminal, write code and create tools it needs to complete tasks.
Give it a longer-running goal, and Muse can monitor progress, react to events and contact you when something changes or it needs a decision.
Muse wants to know you
Muse can remember details from conversations and use them later.
Meta gives the example of Muse turning an Instagram recipe someone saved into a grocery list, suggesting a dinner-party menu and remembering guests' dietary restrictions.
It can connect to services including email, calendars, payments and shopping, with users choosing which services they allow it to access.
That combination of memory, access and autonomy could make Muse useful. It also raises an obvious question: how much would you trust Meta's AI with?
What about passwords and payments?
Meta says Muse operates inside a Muse Secure VM, a dedicated virtual computer in the cloud where the agent and connected data are kept.
A separate AI system called Sentinel oversees Muse's access outside that environment. Meta says it asks users for permission when necessary.
The company also says Muse cannot see users' passwords or payment information. Credentials are stored separately, so the agent can use them without seeing them directly.
For consequential actions, Muse can stop and ask for approval. For example, sending an email or making a purchase can trigger a confirmation screen.
Meta says information stored inside Muse's virtual machine is not shared with its advertising systems, while users can opt out of having their Muse interactions used to train Meta's AI models.
Internal testing reportedly found problems
Those safeguards will face scrutiny.
Reuters reported that Meta employees testing Muse encountered reliability and privacy problems before launch.
Internal posts reviewed by Reuters reportedly described cases in which sensitive information was uploaded without permission and an agent circumvented safeguards to expose personal iCloud photos.
Meta did not respond to Reuters' request for comment on those specific incidents.
Meta Vice President of AI Products Vishal Shah told Reuters the company had previously delayed Muse to improve security. He acknowledged it was impossible to guarantee the agent would never make a mistake, while saying Meta had designed it to be as safe, secure and private as possible.
Meta's trust test
That leaves Meta with a challenge beyond making Muse intelligent.
The company has faced years of scrutiny over privacy and data handling, including a $5 billion US Federal Trade Commission penalty in 2019 over privacy violations.
Muse is effectively asking users whether the convenience of an autonomous AI is worth giving it access to parts of their digital lives.
How much does Muse cost?
Muse is initially rolling out to adults in the US on iOS, Android, the web and through WhatsApp.
Most use will be free, while TechCrunch reports two paid tiers: Power at $20 per month and Maximum at $100 per month.
For now, Muse is US-only, with Meta yet to announce a UAE launch date.
AI chatbots have convinced millions of people to share their questions with artificial intelligence. Muse is asking them to hand over the to-do list — and trust it to get the job done.
