Meta’s Muse is positioned as an AI assistant that can act on your behalf, from sending emails to handling small tasks. The trade-off is access to personal data such as bank accounts, passwords, contact lists, chats, and photos.
The problem is that users may not understand what they are authorizing. The interface also encourages people to share more data than necessary and could pull information about people around them to build profiles without their knowledge or ability to refuse. This turns user convenience into a privacy risk for others as well.
How Far Does Muse Go in Requesting Data About People Around You?
Muse does not learn only about the account owner. It also links information about friends, family, former dating partners, and people the user has contacted to understand the user’s context.
Security researchers found that the system has a process for collecting and updating profiles of everyone in a user’s life on an hourly basis, even if those people have never authorized Muse to collect their data.
When a Personal Assistant Starts Knowing Too Much About Other People
A technology reviewer once allowed Muse to access his data without realizing that the system could reveal his home address and send strangers to his apartment.
The problem does not end with a mistaken setting. Chats, emails, contact lists, and photos may contain information about other people as well. Those people never installed Muse and never consented to having their data shared, yet they have almost no way of knowing whether their privacy has already been used.
Where Muse Fits in Meta’s AI Assistant Landscape
Muse is Meta’s “AI personal assistant,” designed to act on behalf of users—for example, filling out invoices and sending emails—by using surrounding data to help complete tasks before the user gives an instruction.
Sources say Muse became available last month, but it has already been downloaded millions of times and reached number one on the Apple App Store. This reflects how AI assistants like this are moving closer to real-world use in everyday life.
From Smart Glasses to an Assistant That Reads Life Context
| Factor | Muse | Meta’s “pervert glasses” |
|---|---|---|
| Data collection | Emails, messages, contacts, photos, and data about people around the user | The source did not provide details |
| Ability to refuse | The source says there is no meaningful opt-out option | The source describes it similarly |
| Impact on people around the user | Builds profiles of people in the user’s life from digital data | The source did not explain the impact in detail |
This is therefore a comparison of privacy issues, not a comparison between an older and newer model, because the source did not provide information about the previous version or technical product details about the “pervert glasses.”
What Muse Can Do—and What Happens When It Is Used in Practice
Muse uses emails, messages, contact lists, and photos to build context—for example, remembering that the person who sent an invoice was a plumber the user had hired before.
It can also remember personal information, such as which flowers a spouse once said they liked. That sounds convenient, but it also means information about people around the user may be incorporated into profiles without their knowledge.
Regarding reading data from devices, some users reported that Muse pulled large amounts of text they believed it should not have accessed, while Meta explained that Mac users must grant full disk access first.
In another case, Muse revealed a reviewer’s home address and sent a stranger to the residence. The reviewer had previously allowed access to his data but did not realize it would be used in this way.
How Does Muse Compare with Other Options?
The source places Muse among AI assistants that perform tasks autonomously, but does not provide enough information to assess its capabilities, price, or direct competition with other options.
| Factor | Muse | Signal | Not giving data to an AI agent |
|---|---|---|---|
| Type | Autonomous AI assistant | Encrypted messaging app | Data-management approach |
| Information from the source | Some details provided | Mentioned only | No product details |
| Capabilities and price | Insufficient information | Insufficient information | Insufficient information |
| Direct competitor | Cannot be determined | Cannot be determined | Cannot be determined |
The Benefits That Make People Want to Use It—and the Risks They Overlook
Muse can handle routine tasks such as filling out invoices and sending emails, while also taking certain actions before the user asks. This makes it seem like an intuitive assistant that can reduce small, tedious tasks.
But behind the scenes, it requests access to a large amount of data, including bank accounts, passwords, contact lists, chats, and photos. The risk is not limited to the user’s data, because information about friends and people around them may also be included—even though those people never consented.
Pros
- +Helps fill out invoices and send emails
- +Performs certain tasks before the user asks
Cons
- −Collects large amounts of personal data
- −People around the user have no way to opt out
The Price That Is Not Listed in the Storefront
The cost of Muse does not end with convenience. It includes the bank-account data, passwords, contact lists, chat records, emails, and photos that may be exposed to the system.
The risk also extends to address information, as well as data about friends, family, and everyone the user has contacted, even if those people never consented themselves.
Before Entrusting Other People’s Data to an AI Assistant
A user’s consent does not mean that their friends, family, or past contacts have consented to having their own data used as well.
Before enabling Muse, carefully review the permissions granted to the app. Giving an AI access to the context of your life may affect people around you who have no idea that their data is already being used to build an understanding of them.