Preparing only what matters to you
whatmatters.ae

Meta’s New Muse AI Can Read Emails, Make Purchases and Keep Working While You Sleep

Meta’s New Muse AI Can Read Emails, Make Purchases and Keep Working While You Sleep
Meta’s New Muse AI Can Read Emails, Make Purchases and Keep Working While You Sleep

By Thasmiya

Meta has launched **Muse**, a new personal AI agent designed to do much more than answer questions. It can connect to everyday apps, work through tasks in the background, send emails, book travel, fill out forms and even make purchases on a user’s behalf.

The technology marks Meta’s latest move towards what CEO Mark Zuckerberg has described as “personal superintelligence” — AI that understands a person’s context and can actively get things done rather than simply provide instructions.

For now, however, Muse is launching **only in the United States**, so UAE users should not expect immediate access.

## **What exactly is Meta Muse?**

Muse is a personal AI agent powered by Meta’s Muse Spark model.

Instead of having a conversation with an AI and then completing the task yourself, users can give Muse a goal and allow the AI to work through the steps.

For example, you could ask it to organise a holiday. Muse could research travel options, coordinate dates, fill out forms and prepare bookings while you do something else.

For longer tasks, Meta says Muse can continue working even after the user closes the app.

## **Yes, it can read your emails**

Users can connect Muse to services they already use, including email and calendars.

Importantly, Meta says users decide exactly how much access Muse receives. With email, for example, someone could allow it to **read messages only** or additionally give it permission to send emails on their behalf.

This could allow Muse to identify appointments, changes to plans and other information without users manually copying everything into the AI.

Meta says access can be changed or revoked at any time.

## **It can work while you sleep**

One of Muse’s biggest differences from a traditional chatbot is its ability to continue working in the background.

Give it a longer project, close the app and Muse can continue progressing through the task. It returns when something changes or when it needs the user to approve an important action.

Meta gives examples ranging from selling a car and negotiating a bill to adjusting training plans and organising travel.

Essentially, instead of repeatedly prompting the AI, users can give it an objective and allow it to work towards that objective over time.

## **Muse can make purchases too**

This is where the technology becomes particularly interesting — and potentially sensitive.

Muse can shop and proceed to checkout for users. Meta has integrated **Link by Stripe** as a payment option, with Shop Pay planned to follow.

Meta’s New Muse AI Can Read Emails, Make Purchases and Keep Working While You Sleep
Meta’s New Muse AI Can Read Emails, Make Purchases and Keep Working While You Sleep

However, the AI is not supposed to freely spend money without permission.

Meta says Muse asks the user for approval before sensitive actions such as **making a purchase or sending an email**.

Link also generates a one-time-use card for agent purchases, meaning the user’s actual payment-card details are not exposed during the transaction.

## **It can fill forms and negotiate for you**

Muse can operate a browser much like a person.

Meta says it can open websites, fill in forms and even negotiate on a user’s behalf.

That could eventually make it useful for everyday tasks that normally require navigating several websites or repeatedly entering the same information.

Imagine asking an AI to research an item you want, compare options, negotiate where possible, prepare the transaction and return to you when approval is required.

That is the type of workflow Meta is trying to create.

## **It remembers things about you**

Muse also has memory.

It can remember information a user previously mentioned and use that context later.

Meta gives the example of Muse seeing a recipe reel someone saved on Instagram, converting it into a grocery list, suggesting dishes for a dinner party and remembering guests’ dietary restrictions when preparing invitations.

The idea is that users should not need to explain their preferences from scratch every time they ask for help.

Meta says users can tell Muse to forget particular information it has learned.

## **How does Meta plan to keep all of this secure?**

Giving an AI access to emails, calendars, shopping accounts and payments creates obvious privacy and security questions.

Meta has therefore built Muse around what it calls the **Muse Secure VM**.

Each user’s Muse operates inside its own dedicated virtual computer in the cloud. Connected credentials and data are stored within that environment.

A separate AI system called **Sentinel** monitors Muse’s actions. According to Meta, nothing Muse does can reach the internet unless Sentinel approves it or asks the user for permission where necessary.

Meta also says Muse itself cannot see passwords or payment methods stored securely for it to use.

Users receive an audit trail showing what Muse has done and what it intends to do.

## **But there are still security concerns**

The launch is also attracting scrutiny precisely because of how much access an autonomous agent could potentially receive.

Reuters reported that internal testing uncovered security and reliability problems, including concerns around the handling of sensitive personal data. Meta delayed the product from an earlier planned launch while working on security.

Meta’s own researchers acknowledge that allowing software to operate unattended across inboxes, calendars and computer systems creates new safety challenges.

So while Meta is promoting substantial safeguards, giving any autonomous AI access to sensitive accounts requires careful consideration from users.

## **Will your information be used for Meta ads?**

Meta says conversations with Muse and information stored inside its virtual machine **will not be shared with Meta’s advertising systems**.

Users can also choose not to allow their Muse interactions to be used to train Meta’s AI models.

Later in 2026, Meta plans to introduce a **Muse Confidential VM**, where the entire environment — including conversations and personal data — will be encrypted using a key held only by the user. Meta says this would mean even Meta itself could not access that information.

## **Can you use Meta Muse in the UAE?**

**Not yet.**

Muse is initially rolling out in the **United States** on iOS, Android and the web, and users can also interact with it through WhatsApp.

Meta has not announced a confirmed UAE release date.

That distinction is important because UAE residents seeing demonstrations of Muse online may assume it is already part of the Meta AI or WhatsApp experience available locally.

## **How much does Muse cost?**

Meta says most of what people need from Muse will be available for free, with subscription options for users who want greater usage or capabilities.

The company is positioning the service for mainstream consumers rather than only businesses or developers.

## **Is this different from the Meta AI we already know?**

Yes.

Traditional Meta AI features largely focus on answering questions, researching information, generating content and helping users plan.

Meta has already been gradually adding more agent-like capabilities, including connecting with calendars and emails, creating presentations and providing scheduled daily briefings.

Muse takes the concept considerably further.

Instead of simply telling you **how** to accomplish something, the goal is for the AI to actually perform much of the work itself.

## **Why this matters**

Muse represents a bigger shift happening across artificial intelligence.

The first generation of mainstream AI assistants waited for questions and returned answers. The emerging generation of **AI agents** is being designed to take actions.

That means the next question may no longer be:

**“Can AI tell me how to do this?”**

It could increasingly become:

**“Can AI just do this for me?”**

Meta’s Muse can already work in the background, interact with apps, read authorised emails, complete forms, shop and prepare transactions while waiting for approval on sensitive actions.

Whether people will trust an AI with that much access may ultimately be just as important as what the technology can do.

## **Source Credits**

Meta, Meta AI Research, Reuters and Gulf News. Meta’s official Muse announcement provides the product’s capabilities, privacy controls, payment system and US availability, while Reuters independently reported on its launch and internal security concerns.

## **Image Credits / Preview**

For the main article image, Meta’s official launch page contains downloadable Muse product visuals and demonstration media that are preferable to republishing a news publisher’s copyrighted image.

**Image credit:** Meta ⁠Official Meta Muse launch page and media

**Additional reference:** ⁠Meta AI Research — How We Built Safety Into Muse

**Original news reference:** ⁠Gulf News — Meta’s new AI can read your emails, spend your money and work while you sleep

What Matters — UAE news, events and guides