Meta has recently introduced its new personal artificial intelligence assistant, Muse, which swiftly surged to the top of Apple’s App Store charts for free iPhone applications. However, security researchers examining the inner mechanics of the agent discovered that it runs an automated hourly process to construct detailed relationship profiles about people in a user’s life.
These compiled relationship pages contain a vast amount of sensitive personal information, including names, specific relationship connections, recurring discussion topics, important dates, and vital milestones. Because of this sweeping data gathering, individuals can be profiled by the application even if they do not personally download or use the Muse software.
Understanding the Privacy Implications of AI Profiling
How Relationship Mapping Works
Independent security analysts extracted the system prompts and operational instructions of the assistant, revealing that the software actively attempts to map human relationships like a close friend. This mechanism triggers deep data collection routines without explicit consent from the secondary individuals being cataloged in the system.
Meta has defended the feature by asserting that context regarding users and their personal contacts is strictly necessary for the agent to execute daily tasks effectively. The tech corporation maintained that this data is harvested exclusively from public records and inputs provided directly by the user, and is subsequently stored inside secure virtual machines.
Additional Hurdles and Vulnerability Patches
System Security and External Roadblocks
Beyond these mounting privacy concerns, the deployment of the assistant has encountered multiple unexpected technical and commercial hurdles. Notably, Amazon temporarily blocked the application from executing automated shopping tasks on its retail platform.
Furthermore, developers had to rush out a fix for a critical zero-day vulnerability discovered in the Mac version of the software. If left unpatched, this security flaw could have allowed malicious actors to hijack the assistant’s elevated system permissions and compromise user devices.
Here is the source article for this story: Meta’s Muse Is Collecting Information on Everyone in Your Life
