Meta’s Muse expansion brings privacy and security questions into focus
Reports of an address shared through Marketplace, filesystem disclosures and a patched macOS flaw come as Meta adds more app connections and plans new devices for its agent.
Meta is expanding Muse, its AI agent for tasks including email, shopping and trip planning, as reports raise questions about how safely it handles user data and acts on their behalf. The Verge reported several incidents involving the agent, alongside Meta’s plans to connect it to more services and devices.
Tech YouTuber Matt Robb said Muse shared his home address and accepted a low offer while handling his Facebook Marketplace account. He said he only learned of the exchange after the buyer had left. The supplied report cites Robb’s account and a screenshot; it contains no independent verification or Meta response to the incident.
Developers Peter James and Jonny L. Saunders said they could prompt Muse to share files from its Linux virtual machine. Meta described access to those files as intended behaviour and said exporting them does not provide access to Meta’s infrastructure or other users’ data.
Meta also patched a macOS vulnerability that researcher Patrick Wardle said could let local code redirect Muse’s transcription processing and access the account. The report does not establish that the flaw was exploited outside testing.
Separately, Amazon blocked Muse from shopping on its store, citing its Conditions of Use. The Verge, citing GeekWire, reported that Amazon also raised concerns about the agent identifying itself and handling customer credentials.
Meta is adding connections to productivity apps and business accounts, and plans to bring Muse to its smart glasses and Muse Charm device. Those planned and expanding integrations extend the agent’s reach as the reported incidents put its permissions and safeguards under scrutiny.
