Meta is introducing Muse, an AI assistant designed to help users manage routine activities such as online shopping, email management, and travel planning. The company describes Muse as a personal AI agent that can operate independently, launch applications, complete forms, and negotiate on a user's behalf. For extended tasks, Muse continues processing after the app is closed and notifies users if changes occur or approval is needed for actions like purchases.

The service will launch in the United States across iOS, Android, and the muse.ai platform, with AI-powered glasses expected in the future. Access will be free for most users, though Meta indicated paid tiers will exist for advanced capabilities without disclosing pricing or feature details.

Meta positions Muse as accessible to non-technical users, requiring no setup beyond messaging through the app or WhatsApp. The assistant retains context from prior conversations to offer proactive suggestions. Privacy measures include cloud-based virtual machines that isolate user data, a separate monitoring agent called Sentinel to prevent unauthorized internet access, and user controls to opt out of data collection for model training or request deletion of specific memories.

A more secure, encrypted version of the virtual environment is planned for later this year. Integration with 1Password and Shop Pay is underway, and checkout functionality will use Stripe’s Link system. Muse runs on Meta’s proprietary Muse Spark model and represents the company’s latest effort to compete in the consumer AI space.

The rollout occurs amid ongoing challenges for Meta in rebuilding trust following previous controversies around social media impacts and data privacy. The company has previously faced issues with AI-related policy violations, including an AI-generated news feed and a chatbot vulnerability that compromised Instagram accounts. Muse aims to leverage Meta’s large user base while addressing security concerns through technical isolation and user-controlled data handling.