🔍 Read the full analysis: Meta’s AI Agent And Your Messages: A Question Of Access on ThorstenMeyerAI.com
Get the latest gadgets 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
Meta’s Muse personal AI agent can access data through connectors, including certain messages when a user connects or shares access to those services. WIRED and TIME reported that Muse’s instructions direct it to maintain profiles of people in users’ lives; Meta did not dispute TIME’s findings and says each user controls the agent’s access. The scope of non-users’ ability to review or remove information about themselves remains unclear.
Meta announced the expansion of its Muse AI agent to Singapore on October 27, as reports raised questions about what the assistant can learn from messages and other information its users connect. Meta says users control what access Muse receives, but WIRED and TIME reported that its instructions direct it to keep profiles of people mentioned in users’ communications, including people who have never signed up.
Muse works through connectors that let it use information from other services to carry out tasks. Meta’s help materials say users can connect email, calendars and Mac Messages; access to Messages requires permission in macOS and activation of the connector. Meta says email access lets Muse search messages but does not download an entire inbox unless the user asks it to. Its materials also say Facebook, Instagram and Threads connect automatically when accounts share an Accounts Center with Muse.
Meta chief AI officer Alexandr Wang said the available connectors include Instagram direct messages and Messenger. The source material says Meta’s public help pages do not document the full scope of what those two connectors can read. Meta describes Muse as running in a private virtual machine for each user. In WhatsApp, the source says, Muse is a chat that a user opens, rather than an agent that reads their other chats by default.
WIRED reported on instructions obtained by security researcher Karan Joshi that direct Muse to create a page for every person in a user’s life. TIME later reviewed instructions that it said tell Muse to update records hourly about the user and people named in chats, messages and emails, including details about relationships and disputes. Meta did not dispute TIME’s findings. A company spokesperson told TIME: “Each person stays in control of their Muse and decides how much access it gets.”
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 Other People’s Messages Become Data
The issue extends beyond whether a Muse user wants an assistant to summarize an inbox. Messages contain information about both the sender and recipient, so an agent permitted to search one person’s communications may process details about people who never chose to use Meta’s product. The reports suggest that information could be organized into persistent profiles, rather than used only to answer a single request.
That matters for people who communicate with Muse users about sensitive subjects, including family matters, workplace disagreements or health concerns. The source material does not establish that every message is continuously read, or that every person mentioned receives a profile. But reports of hourly profile updates, combined with Meta’s stated collection of information “about you or others,” make the boundaries of consent and retention central questions for users and non-users alike.
The commercial setting adds another reason to watch how the system develops. Meta sold $25 billion in stock at $625 a share and $12.5 billion in preferred stock, while Nvidia said it would invest $3 billion as the companies expand their work on Muse. The source material says Meta will deploy Nvidia chips for the agent. Those financial and infrastructure commitments indicate the companies are investing in the service’s growth; they do not, by themselves, establish how Muse handles any particular person’s data.
Products Worth Considering
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
Bose QuietComfort Wireless Bluetooth Headphones with Active Noise Cancellation, Mic, USB‑C Charging, Deep Bass, 24‑Hour Battery – White Smok
Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Black
As an affiliate, we earn on qualifying purchases.
Muse’s Rollout and Data Connections
Meta launched Muse in the United States on September 8, 2026, according to the source material. Japan followed on October 23, described there as its first market outside North America, and Meta announced Singapore on October 27. TIME reported more than 5 million downloads within weeks; the source does not provide a precise measurement window or comparison baseline for that figure.
The assistant is designed to act on a user’s behalf, including tasks such as answering email, booking tables and shopping. Its access depends on product settings and connectors, although Meta’s materials also describe some connector data as being shared proactively. The source says training on users’ data is on by default and that Muse users can disconnect connectors, change training settings, edit a memory file or reset the agent. Custom connectors are also possible, and Meta’s help page warns that the company does not review them.
The reports about profiles are distinct from Meta’s general product descriptions. WIRED reported on instructions obtained by Joshi; TIME later examined instructions separately. The source says Meta did not dispute TIME’s findings, but does not provide a detailed company explanation of how profiles are created, stored, or applied in individual cases.
““They’re trying to know you like a friend, which is honestly pretty creepy.””
— Karan Joshi, security researcher, speaking to WIRED
Consent, Retention and Profile Controls
Several practical details remain unresolved. The source material does not establish exactly which messages or fields Muse reads in every connector, how often data is accessed for different tasks, or how long information about people mentioned in communications is retained. It also does not establish whether the reported hourly updates apply to every Muse user or every person named in their messages.
Meta’s help material says Muse may remember information it learned even after a user deletes the original item, and describes forgetting as working “to the best of its ability.” That leaves a distinction between removing a source message and removing information inferred or retained by the agent. The source says TIME reported that Muse’s instructions discourage it from pointing out that original messages may remain visible; the specific behavior and its frequency are not independently detailed here.
It is also unclear what controls people who do not use Muse have over profiles or summaries created from another person’s data. The source identifies WhatsApp’s Advanced Chat Privacy as one way a non-user may limit what is retained in other people’s archives, but gives no equivalent universal control across email, Instagram, Messenger or other connected services.
Meta’s Next Privacy Disclosures
Meta’s Singapore rollout puts the product in another market, but the source provides no timetable for further expansion or for changes to its privacy documentation. The next useful developments would be clearer descriptions of what each connector can access, how profile information is stored and deleted, and whether people mentioned in a user’s records can object or request removal.
For now, Muse users can review connected services, disconnect connectors they do not need, check model-training settings, edit the memory file or reset the agent. People who do not use Muse have fewer direct options, according to the source. Whether Meta will add controls aimed at those non-users—and how the company will address the reported profile instructions—remains unanswered.
Key Questions
Can Meta’s Muse read Instagram direct messages?
Meta chief AI officer Alexandr Wang said Muse has an Instagram connector that includes DMs. The source material says Meta’s public help pages do not spell out the full scope of that access.
Does Muse automatically read all of a user’s email?
Meta’s help materials say connecting email lets Muse search it, but it will not download the entire inbox unless the user asks it to. The source also says connector access may vary by service and task.
What did reports say Muse does with information about other people?
WIRED and TIME reported that Muse’s instructions direct it to keep pages or files about people in a user’s life and update information drawn from communications. Meta did not dispute TIME’s findings; the source does not establish that every person mentioned is profiled.
Can someone who does not use Muse delete information about themselves?
The source does not identify a general removal mechanism for non-users. It says people may have limited options, including WhatsApp’s Advanced Chat Privacy to reduce what is retained in other people’s archives, but that does not address every service.
What can a Muse user do to limit access?
According to the source material, users can disconnect connectors, switch off model training, edit Muse’s memory file or reset the agent. Users should also review any custom connectors, which Meta’s help page says it does not review.
Source: ThorstenMeyerAI.com
