Sources: OpenAI is developing features to counter SpaceX's Grok Bot "teammates", and has discussed creating a personal AI assistant to compete with Meta's Muse

Meta's AI assistant proactively requests access to your financial and communication data, nudging you to link accounts for enhanced personalization.

Meta has launched Muse, a new personal AI assistant that automates everyday tasks like booking reservations and managing inboxes. The app, which saw over 900,000 downloads in its first week, allows users to connect personal data sources like email and bank accounts. Muse browses the internet via a "virtual machine" to complete tasks, and integrates with Meta's platforms including WhatsApp and Messenger. While Meta claims Muse has built-in protections and user controls, critics express concerns about its extensive data collection practices. Users are automatically opted into AI model training, though this can be disabled. The company plans to release "confidential" versions of its virtual machine later this year that prevent Meta from accessing user data. Despite assurances, privacy advocates view the opt-out data collection as a significant red flag, citing Meta's history with user trust.

Meta's Muse represents a significant push into mainstream personal AI agents, aiming to automate tasks by deeply integrating with user data and online services. The agent's "virtual machine" browsing capability allows it to perform complex actions like online shopping and reservations, raising questions about user autonomy and the potential for automated decision-making to shape personal preferences over time. Meta's strategy relies heavily on its existing platform ecosystem to drive adoption and data collection, positioning Muse as a central hub for personal organization and digital interaction.

The extensive data collection by Muse, including opt-in AI model training and a "Memory" document storing user interactions, has drawn criticism from privacy advocates. Concerns are amplified by Meta's past data handling practices and the automatic opt-in for training, which requires users to actively disable it. This approach, coupled with potential revenue opportunities beyond advertising, suggests a business model where user data fuels AI development and potentially influences the information presented to users, impacting their digital experience and purchasing behavior.