Read the full analysis: Don’t Use Muse? Check What Meta’s AI Agent Can See on ThorstenMeyerAI.com
TL;DR
Meta’s Muse personal AI agent can access information through connected email, calendar, messaging and Meta accounts, subject to permissions and settings. Reports say its internal instructions direct it to maintain regularly updated pages about people mentioned in a user’s communications, including people who never signed up; Meta did not dispute those findings and says each Muse user controls the agent’s access.
Meta’s Muse personal AI agent can access information from accounts its users connect, and reports by WIRED and TIME say its instructions direct it to maintain updated pages about people mentioned in users’ chats, messages and emails. That means people who have never used Muse may still be described in an agent’s records; Meta did not dispute the reports and said each user controls what the agent can access.
Muse uses connectors to interact with other services. Meta’s help pages say connecting email lets the agent search messages but does not download the entire inbox unless the user asks. Calendar connectors may send updates to Muse proactively. Meta also says Facebook, Instagram and Threads connect automatically when the accounts are in the same Accounts Center.
Meta chief AI officer Alexandr Wang said the agent has connectors for Instagram, including direct messages, Threads, Facebook, including Marketplace, and Messenger. The scope of those messaging connectors is not fully detailed on Meta’s help pages. Access to Mac Messages requires the user to grant macOS Full Disk Access and enable the connector. Meta warns that it does not review custom connectors that users create.
In an October 3 report, WIRED described instructions for Muse to create “a page for every person in the user’s life.” TIME reported three days later that the agent updates files hourly and can record details about people named in chats, messages and emails, including relationships and disputes. TIME said agents also share de-identified lessons across the system. Meta did not dispute the findings. Its spokesperson said each user controls their Muse and decides how much access it gets.
Muse · as of 28 Oct 2026
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.
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 followedNotes on how people met, shared interests, disputes, “tensions and alliances”People who never signed up are mapped by other users’ agentsDe-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 itConversations 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 Records
The privacy issue extends beyond the person who installs the agent. When a user gives Muse access to communications, those messages can contain details about friends, relatives, colleagues and other contacts. The reports describe Muse organizing some of that information into persistent profiles, rather than treating each message as an isolated request.
That creates a gap between the user’s control over permissions and the privacy expectations of people whose words appear in the user’s accounts. A person who never signed up may have no direct Muse settings to change, even if their messages are among the material an agent can search or summarize. The reporting raises questions about consent, retention and how profiles inferred from conversations are handled.
The commercial stakes are also substantial. The source report says Meta sold $25 billion in stock at $625 per share and $12.5 billion in preferred stock, while Nvidia, whose chips Meta plans to use for Muse, is investing $3 billion. That financial relationship does not establish how Muse handles data, but it underscores the scale of the infrastructure and business ambitions behind the product.
Muse’s Launch And Growing Reach
Meta describes Muse as a personal AI agent that runs on a cloud computer and can perform tasks such as answering email, booking tables and shopping. The company says each user gets a “Muse Secure VM,” a private virtual machine that other users’ agents cannot access. Users can interact with Muse in its app, on a Mac or through a chat opened directly in WhatsApp.
Meta launched Muse in the United States on September 8, 2026, followed by Japan on October 23 and Singapore on October 27. TIME reported more than five million downloads within weeks. Meta called the product “the world’s first personal AI agent built for everyone,” a company characterization rather than an independently established ranking.
Meta’s policy says Muse may store information about the user and other people. The source report also says model training is on by default, and describes user controls including disconnecting connectors, changing the memory file, resetting Muse and turning off training. Meta’s help material cautions that asking Muse to forget information is not necessarily the same as erasing every trace of it.
What The Reports Do Not Establish
The available reporting does not establish how often Muse actually creates or updates these profiles, how long each profile is retained, or how users can inspect and delete every record about another person. The instructions described by WIRED and TIME show what Muse was directed to do; they do not, by themselves, document the contents of every user’s agent or prove that every communication is processed in the same way.
Meta’s help pages and policy, as summarized in the source report, describe different kinds of access, including task-based searches and proactive sharing by some connectors. The exact data fields, timing and retention rules for Instagram DMs and Messenger remain unclear. Meta also says its virtual machines are private from other users’ agents, but that claim does not answer every question about Meta’s own handling of data.
The source report says Meta’s help pages describe “forget” as working “to the best of its ability” and warn that Muse may still remember information it learned from deleted material. The full deletion process and the practical effect of turning off training or resetting the agent are not detailed here. Nonusers’ options also vary by app and may not remove information already available in another person’s account.
Controls Users Can Change Now
Muse users can review connected services and disconnect connectors they do not want the agent to access. They can also check the model-training setting, edit the memory file or reset the agent, according to the controls described in the source report. Users should consult Meta’s current settings and help pages for the exact steps, which may change as the product expands.
People who do not use Muse have fewer direct controls because the agent is attached to someone else’s account. The report points to WhatsApp Advanced Chat Privacy as one setting that may limit what can be taken from a chat, but it does not establish that this removes information already stored elsewhere or applies to other services. Meta’s next clarifications on data retention, deletion and messaging connectors will determine how much users and their contacts can control in practice.
Key Questions
Can Muse access messages from people who do not use it?
Potentially, if those messages are in an account or service the Muse user has connected and the agent can access. Reports say Muse’s instructions direct it to keep pages about people mentioned in communications. The reporting does not show that every nonuser’s messages are accessed.
Does Muse automatically read every connected message?
Meta’s help information says some access happens when a task requires it, while its privacy policy describes connectors that can share data proactively. The exact behavior depends on the connector and settings; the available source material does not establish that Muse reads every message in every connected account.
What can a Muse user do to limit access?
Users can review and disconnect connectors, check the model-training setting, edit the memory file or reset Muse, according to the source report. Meta’s help pages should be checked for current controls and their effects.
Can a nonuser delete a profile Muse has about them?
The available reporting does not identify a direct control for nonusers to inspect or delete a profile held by another person’s agent. Options such as WhatsApp Advanced Chat Privacy may limit what can be shared from a chat, but are not shown to erase information already stored.
What did Meta say about the reported profiling instructions?
Meta did not dispute the findings reported by WIRED and TIME. A spokesperson said each user controls their Muse and decides how much access it gets. The source material does not provide a detailed company response on retention or deletion of profiles about nonusers.
Source: ThorstenMeyerAI.com