🔍 Read the full analysis: What Meta’s AI Agent Sees When You Don’t Use Muse on ThorstenMeyerAI.com
Get the little things that make your day delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Reports by WIRED and TIME say Meta’s Muse agent is instructed to maintain profiles of people mentioned in users’ messages and update them hourly. Meta did not dispute the findings, saying users control Muse’s access; the scope of data use and deletion remains unclear, and people who never signed up have few direct controls.
Meta’s Muse AI agent may build and regularly update profiles of people who never signed up, using messages and other information available through a user’s connected accounts, according to investigations by WIRED and TIME. Meta did not dispute the reported instructions, but said each Muse user controls what the agent can access—a distinction that leaves people discussed in those users’ messages with few direct ways to limit the processing.
Muse is a personal AI agent that Meta says can act on a user’s behalf, including searching email and helping with tasks. It uses connectors to access selected services. Meta’s help information says email access lets Muse search messages without downloading an entire inbox unless the user directs it to do so. Calendar connections can send updates to the agent. Meta also says Facebook, Instagram and Threads accounts in the same Accounts Center connect automatically. Meta chief AI officer Alexandr Wang said the Instagram connector includes direct messages and that Messenger is also supported.
In an October 3 report, WIRED described instructions that security researcher Karan Joshi obtained by asking Muse about its files. The instructions reportedly directed the agent to create “a page for every person in the user’s life.” TIME separately reviewed instructions and reported that Muse updates information hourly about the user and people mentioned in chats, messages and emails. The reported profiles can include relationships, shared interests and tensions. Meta did not dispute those findings in the source material.
Meta’s privacy policy says Muse can store “information about you or others.” Its help materials also caution that deleting information does not necessarily remove what Muse has learned from it: the agent may still remember deleted material. The source reports that Meta’s instructions tell Muse not to point out that original messages may remain visible. That account is attributed to TIME; the precise retention behavior and its application to different connected services are not fully specified in the material provided.
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
How Other People Enter Muse’s Memory
The reported profiling raises a privacy issue that extends beyond the people who choose to use the product. A person’s messages can describe relatives, colleagues and friends, and access granted by one account holder may allow an agent to process information about those contacts. They do not necessarily receive a notice or a setting in Muse to manage that use.
The distinction matters because Meta’s statement focuses on the user’s control over connectors, while the reported files concern people who appear in connected information. Users can disconnect services, change settings or edit their own agent’s memory. Non-users have fewer direct controls; the source points to options such as WhatsApp’s Advanced Chat Privacy as a way to limit what is shared in other people’s archives, not as a way to erase information already processed by Muse.
The commercial context is also relevant, though it does not establish a link between financing and any particular privacy decision. Meta sold $25 billion in stock at $625 a share and $12.5 billion in preferred stock, according to the source. Nvidia, whose chips Meta plans to deploy for Muse, is investing $3 billion. The scale of the investment underlines the business importance of the agent, while the source’s claim that its business case depends on deep access to personal data is an interpretation, not a disclosed Meta finding.
Products Worth Considering
Ring Outdoor Cam 4K Plus Battery Security Camera, 3-Pack, White
Ring Floodlight Cam 2K Wired Security Camera, 3-Pack, Black
Dyson Purifier Hot+Cool HP1
As an affiliate, we earn on qualifying purchases.
Muse’s Connectors and Rollout
Meta launched Muse in the United States on September 8, 2026, and announced Japan as its first market outside North America on October 23, followed by Singapore on October 27. Meta describes Muse as a cloud-based personal agent operating through a private virtual machine. It can be used in its app, on a Mac or in a WhatsApp chat. Meta says a WhatsApp conversation with Muse is a separate chat, rather than access to a user’s other WhatsApp conversations.
Access depends on the connector and permissions. On a Mac, the source says Muse can connect to Apple Messages only if the user grants macOS Full Disk Access and enables the connector. Meta says users can also create custom connectors, but warns that it does not review them. Its privacy materials describe some connectors as sharing information proactively, while also saying Muse generally accesses information when a task requires it. The source does not establish that every connector continuously reads all available content.
TIME reported that Muse had more than 5 million downloads within weeks of launch. Meta’s help pages describe controls for users, including disconnecting connectors, turning off model training, editing memory and resetting the agent. The source says model training is on by default, but provides no detail here on how that setting applies to each category of data.
“They’re trying to know you like a friend, which is honestly pretty creepy.”
— Karan Joshi, security researcher, quoted by WIRED
What the Reports Do Not Establish
The source material does not specify exactly which details Muse retains for each person, how long those profiles persist, or whether every connected service supplies information in the same way. Meta’s public help pages reportedly do not spell out what its Instagram and Messenger connectors read, despite Wang’s statement about their availability. It is also unclear how often the agent accesses data outside the reported hourly updates to its files.
The reports describe instructions and behavior observed in particular investigations; they do not establish that every Muse user’s agent has created a profile on every contact, or that every message is processed. The material also does not explain in detail how de-identified lessons are separated from identifiable information, how users’ model-training settings affect these data, or what deletion means for copies and derived memories. Meta’s general assurance of user control does not answer those operational questions.
Details Needed on Data Controls
The next useful development would be clearer, service-by-service information from Meta about what Muse can access, when it accesses it, and whether connected information is used to create or refresh profiles of people other than the user. Readers will also need specific explanations of retention, deletion, model training and the reported sharing of de-identified lessons.
For now, Muse users can review connected accounts, disconnect connectors they do not need, change model-training settings, edit memory or reset the agent, according to the source. People who do not use Muse may be able to limit what others can retain or share through settings in particular apps, but the material does not identify a general process for non-users to inspect or remove information already handled by someone else’s agent.
Key Questions
Can Muse create information about people who do not use it?
WIRED and TIME reported that Muse instructions direct it to keep profiles of people in a user’s life and update them using information in messages and other connected data. The reports do not establish that every non-user has been profiled.
What information can Muse access?
Access depends on the user’s connectors and permissions. Meta lists email, calendar and its own services; Mac Messages requires additional permission. The exact scope of some social messaging connectors is not fully documented in the supplied material.
Does deleting a message erase what Muse learned from it?
Not necessarily. Meta’s help information says Muse may still remember information learned from deleted material. The source does not give a complete account of how this works across different services or memories.
What can a non-user do to limit this?
The source says non-users have fewer direct options than Muse users. It points to app-level privacy settings, including WhatsApp’s Advanced Chat Privacy, as ways to limit what may enter other people’s archives; it does not describe a general opt-out from Muse profiling.
What has Meta said about the reports?
Meta did not dispute the reported findings in the source material. A spokesperson said each user controls their Muse and decides how much access it receives; the statement does not directly explain controls for people whose information appears in a user’s connected data.
Source: ThorstenMeyerAI.com
