Meta's AI assistant prioritizes data collection over task automation

Meta's AI assistant, Muse, has gained traction with over 900,000 downloads in its first week, offering users the ability to automate tasks like booking reservations and managing emails. The app, available on Instagram, WhatsApp, and Messenger, uses a virtual machine to browse the internet on users' behalf, demonstrating improved reliability compared to previous AI tools. Critics argue that Muse's primary function is data collection, with features like 'Memory' storing user preferences and interactions without an opt-out toggle. Meta claims data is 'sanitized' for AI training, but consumer advocates warn of potential privacy risks and manipulative tactics. Rory Mir of the Electronic Frontier Foundation highlights how interactions with Muse effectively feed data to Meta. Despite Meta's assurances of 'confidential' versions later this year, concerns persist about the app's impact on user autonomy and decision-making. The author ultimately deleted Muse, citing fears of over-reliance on AI and loss of personal agency. Meta's Tarek Sheasha defends data collection as essential for improving AI models, while Calli Schroeder of the Electronic Privacy Information Center criticizes the company's approach as a repeat of past mistakes. The app's ability to influence ad targeting through web-tracking algorithms raises further concerns about Meta's data-driven business model. Margaret Mitchell of Hugging Face warns that AI agents may degrade cognitive abilities by reducing user oversight. Muse's integration with Facebook Marketplace and its push for deeper data access underscore its role in shaping user behavior and preferences. The app's persistent prompts for data sharing reflect a broader trend of AI tools prioritizing personalization over privacy.