Meta Denies AI Assistant Muse Accessed Private Messages Without Permission

Meta has denied allegations that its AI assistant, Muse, accessed users' private messages without proper authorization. The claims emerged following a report by Jason Aten, a columnist for "Inc." magazine, who detailed an experience where he alleged Muse gained access to his messages. Meta, however, has stated that the company does not believe its products have accessed messages without user consent.

Andy Stone, Meta's Vice President of Public Relations, clarified on X (formerly Twitter) that the message integration feature within the Muse application on Macs is entirely opt-in. Users must enable full disk access and then connect the messaging connector for Muse to read message content. Stone emphasized that without these multi-layered permissions, Muse cannot access user messages.

The controversy comes amid a history of data handling issues and regulatory scrutiny for Meta. A jury in New Mexico recently found the tech giant misled customers regarding its data practices, stemming from the 2018 Cambridge Analytica scandal. These past incidents contribute to user skepticism, even as Muse remains a popular application topping app store charts.

Previously, David Singleton, head of Meta's Superintelligent Labs, explained the technical safeguards on Threads. He detailed that Muse requires several distinct user-granted permissions and that macOS has built-in protections against unauthorized access. Aten, however, claimed Muse read his messages while full disk access was disabled, suggesting it was synchronizing device notifications. Singleton disputed this, attributing it to an AI misinterpretation of the situation.