🔍 Read the full analysis: What To Know About Meta’s AI Agent, Muse, And Message Access on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Meta’s Muse personal AI agent can use information from connected services, including some messages, to carry out tasks for its user. Reporting by WIRED and TIME says its instructions also tell it to maintain and update profiles of people mentioned in users’ communications; Meta did not dispute those findings. The exact scope of message access, how long information persists, and what control non-users have remain unclear.
Meta announced that its Muse personal AI agent is coming to Singapore on October 27, extending its availability beyond North America as scrutiny grows over the service’s access to connected messages and its reported records about people who do not use it. Meta says users control which services they connect; reporting by WIRED and TIME says Muse’s instructions direct it to keep and update profiles of people mentioned in those communications.
Muse is an AI assistant that runs on a computer in the cloud and can perform tasks for its user, such as answering email, booking tables and shopping. Meta launched it in the United States on September 8, then expanded to Japan on October 23 and announced Singapore on October 27. Meta describes each user’s environment as a Muse Secure VM, a private virtual machine that it says other users’ agents cannot access.
The agent’s access depends on its connectors, which link it to other services. Meta’s help information says email can be searched when connected, but the entire inbox is not downloaded unless the user asks. Calendar data may be sent to Muse proactively. Meta also says Facebook, Instagram and Threads connect automatically when accounts are grouped in the same Accounts Center. Meta chief AI officer Alexandr Wang said the Instagram connector includes direct messages and that Muse has a Messenger connector; the available help material does not spell out the full scope of those connections.
Access to Apple’s Messages app on a Mac requires the user to grant macOS Full Disk Access and enable the connector, according to Meta. Custom connectors can also be created, but Meta warns that it does not review them. Muse is available through its app and on Mac; Meta says users can also open it as a chat in WhatsApp. That does not, by itself, mean Muse reads a user’s other WhatsApp conversations.
Infographic 01 · What Muse can read
Muse reads what its user can read. That includes you.
Connectors decide what the agent sees. The user switches them on. The people in that user’s chats and inbox are never asked.
Sources a Muse user can open to the agent
The agent reads what its user can read. Everyone who writes to that user is inside.
| Source | How it connects | Status |
|---|---|---|
| Email (Gmail, Outlook) | Opt-in connector. Searches when a task needs it; downloads the whole inbox only if told to. | Meta document |
| Calendar | Opt-in connector. Can push changes to Muse on its own (“proactive”). | Meta document |
| Facebook, Instagram, Threads | Connected automatically if the accounts share an Accounts Center with Muse. | Meta document |
| Instagram DMs, Messenger | Named as Muse-only connectors (“Instagram (incl dm’s) … and Messenger”). What they read is not documented. | executive post |
| Mac Messages (iMessage) | Only with macOS Full Disk Access plus the Messages connector, Meta says. One columnist says it synced anyway. | disputed case |
| A place to chat with Muse, not a connector. Test-build text mentions Muse access to end-to-end encrypted chats. | leaked code | |
| Custom connectors | Muse can build links to other services. Meta does not review them. | Meta document |
What the instructions say
Extracted by researchers; reported by WIRED and TIME reported
- “A page for every person in the user’s life”: family, partners, friends, colleagues, people followed
- Notes on how people met, shared interests, disputes, “tensions and alliances”
- People who never signed up are mapped by other users’ agents
- De-identified “lessons” shared across agents; Meta did not dispute TIME’s findings
What Meta says
Privacy policy, help pages, spokespeople Meta document
- Each user’s VM is isolated; no other user’s agent can reach it
- Conversations and VM data are not shared with Meta’s ad systems
- “Each person stays in control of their Muse and decides how much access it gets”
- Forgetting works “to the best of its ability”; deleted data may still be remembered
Sources: Meta Help Center: Connectors; Muse Privacy Policy; Alexandr Wang on X (8 Sep 2026); TIME (6 Oct 2026); Tom’s Guide on WIRED’s report (6 Oct 2026); TechCrunch (30 Sep 2026); Digit (16 Oct 2026). Status labels: Meta document = stated by Meta; executive post = Meta executive on X, not in help pages; disputed = claim and denial on record; leaked = unreleased code, may never ship.
Powered by Thorsten Meyer AI
When One User Connects a Conversation
The privacy issue reaches beyond the person who signs up. If an agent can read a user’s messages, its work may involve information about friends, family members, colleagues and other people who never gave Meta permission to use Muse on their behalf. The practical question is not only what the user asks the agent to do, but also what information from other people’s communications may be retained or used along the way.
That matters especially in shared conversations. A person who does not use Muse cannot directly manage another person’s connectors or memory files. Meta’s controls, as described in the source material, mainly give the Muse user choices about access, training and saved information. For people appearing in someone else’s messages, the available options are more limited and depend on the app and its settings.
The commercial stakes are also notable, though they do not establish how Muse handles data. The source reports a Meta stock sale of $25 billion at $625 a share, alongside $12.5 billion in preferred stock, and says Nvidia is investing $3 billion as the companies expand their work. Nvidia supplies chips Meta plans to deploy for Muse. The funding and infrastructure help explain the scale of the project; they do not, on their own, confirm any particular data practice.
Popular right now
DeLonghi Rivelia Automatic Espresso Machine Bundle with Grinder and Frother, Beige
Ninja Foodi 10-Quart DualZone 6-in-1 Air Fryer, Black, DZ302
LG C2 Series 65-Inch OLED evo 4K Smart TV (OLED65C2PUA, 2022)
As an affiliate, we earn on qualifying purchases.
Connectors, Memory and Rollout
Meta’s privacy information says Muse may collect messages, connector data and “information about you or others.” The help pages distinguish between data an agent searches when a task calls for it and connectors that may share data proactively. The source material does not provide a complete, service-by-service account of what is accessed, when it is accessed or how long it is kept.
On October 3, WIRED reported on Muse instructions obtained by security researcher Karan Joshi. The report said the agent was directed to create “a page for every person in the user’s life,” covering people such as relatives, friends and colleagues. TIME reviewed the instructions separately three days later and reported that the agent updates files hourly about the user and people named in chats, messages and email. Those files reportedly can include relationships, shared interests and conflicts.
The source says Meta did not dispute TIME’s findings. Meta’s spokesperson responded that each person controls their own Muse and decides how much access it gets. That statement describes the account holder’s control; it does not explain what choices are available to people whose information appears in another user’s messages. The source also reports that TIME found instructions for sharing de-identified lessons across the system, but provides no further detail on what those lessons contain or how they are used.
The Limits of Access and Deletion
The reporting raises concerns but leaves important operational details unresolved. The source does not establish exactly which Instagram direct messages or Messenger content Muse can access, whether all connected data is stored, or what prompts or tasks cause it to retrieve particular messages. Meta’s help pages, as described in the source, do not provide a full account of those boundaries.
How long information persists is also unclear. Meta’s help material reportedly says Muse may still remember information learned from content a user deletes, while saying it forgets “to the best of its ability.” The supplied source excerpt cuts off before the full deletion explanation. It is not clear from the material how a user can verify removal, or what happens to reported profiles of other people after a connector is disconnected or a memory is reset.
The source does not include a detailed response from Meta addressing the reported hourly updates or the reported cross-system sharing of de-identified lessons. It also does not establish whether the instructions reported by WIRED and TIME apply in the same way across every country, connector or version of Muse. Those details should not be treated as settled.
Watch for Clearer Privacy Controls
Meta has announced Singapore as its next Muse market, but the source gives no date for general availability there and no schedule for further expansion. The key developments to watch are whether Meta publishes more detailed explanations of each connector, clarifies what message content is retained, and explains how deleting information affects memories derived from it.
For current users, Meta’s described controls include disconnecting services, turning off model training, editing a memory file and resetting Muse. The source says model training is on by default, but does not include the exact steps or explain how these controls differ by app or location. People who do not use Muse have fewer direct options; the source points to WhatsApp’s Advanced Chat Privacy as one way to limit what may be retained in other people’s archives, without explaining the feature’s full reach.
Any assessment should distinguish Meta’s documented connector behavior from Wang’s public description and from the reporting about Muse’s internal instructions. More detail from Meta on access, retention and non-user protections would help clarify how the agent works in practice.
Key Questions
Does Muse automatically read every message on a connected account?
The source does not establish that it reads every message. Meta says connected email can be searched without downloading the whole inbox unless the user asks, while some connectors may share data proactively. The precise scope for Instagram DMs and Messenger is not fully documented in the material.
Can Muse make profiles of people who have not signed up?
WIRED and TIME reported that Muse’s instructions direct it to create and update files about people mentioned in a user’s communications. Meta did not dispute TIME’s findings, but the source does not independently establish how broadly this happens in everyday use.
Can a non-user delete information held by someone else’s Muse?
The source describes controls for Muse account holders, including disconnecting connectors, editing memory and resetting the agent. It does not identify a direct way for a non-user to control or delete information in another person’s Muse archive.
Does deleting a message erase what Muse learned from it?
Not necessarily, according to the help-page wording quoted in the source: Muse may still remember information it learned from deleted content. The supplied excerpt does not include the full policy or explain how a user can confirm that derived information has been removed.
Source: ThorstenMeyerAI.com
