Meta's Muse: Helpful AI or Data Hoover?

Meta's AI assistant, Muse, has seen significant early adoption with over 900,000 downloads in its first week. The app, available on Instagram and WhatsApp, promises to automate everyday tasks. However, users report that Muse frequently prompts them to link additional personal data sources, such as email and bank accounts, for improved functionality.

While Muse demonstrates impressive capabilities, like navigating websites to order items, its design heavily emphasizes data collection. Experts caution that this constant push for more information, coupled with the agent's automation, could lead to users becoming overly reliant and experiencing cognitive degradation. Meta states that user interactions are used for AI model training, which can be opted out of, but this default setting has drawn criticism.

The app's 'Memory' feature stores user interaction details, with options to wipe data but no toggle to disable memory collection entirely. Privacy advocates express concern that the AI's actions, even for tasks like making reservations, could indirectly influence advertising seen by users, aligning with Meta's core business model of data collection and targeted ads.