Meta has launched Muse, a new personal AI assistant designed to help users with daily tasks and projects, as part of its effort to stay competitive in the growing artificial intelligence market. Available in the U.S. through iOS, Android, and the website muse.ai, Muse is described as the first AI agent built for the general public. It can perform a variety of tasks on behalf of users, such as online shopping, drafting emails, planning trips, and even filling out forms. For more complex tasks, Muse can continue working in the background after the app is closed, alerting users if action is needed or if something changes. Muse is powered by Meta's in-house AI model called Muse Spark and is designed to connect with various apps and services, including Meta's own platforms and third-party tools like Google Workspace, Spotify, and Apple Health. Users can link these services one at a time, with more options expected in the future. If a service doesn’t have a public API, Muse can still access it through a browser. The assistant will be free for most users, though two paid plans will be available: Power at $20 per month and Maximum at $100 per month. However, Meta has not yet detailed the specific features of these plans or the limitations of the free version. Meta emphasizes that Muse is easy to use, requiring no technical skills or learning curve. Users can interact with Muse through the app or via WhatsApp, and the assistant can remember details shared with it, allowing it to make more personalized suggestions. Privacy and security are central to Muse’s design, with the agent running on a secure, isolated virtual machine that keeps user data separate from other users. Meta claims that Muse does not share conversations or data with its advertising systems. A separate "Sentinel" agent is also used to block unauthorized internet access. Later this year, Meta plans to introduce a more secure "confidential" version of the virtual machine, encrypted so that even Meta cannot access the data. Additional features like support for 1Password and Shop Pay are also coming. Meta has struggled to rebuild consumer trust after past privacy issues, such as the Cambridge Analytica scandal and recent lawsuits over the harms of social media. The company has provided technical documentation and security measures to address these concerns. Users can choose not to have their interactions with Muse used for AI training and can ask Muse to "forget" specific information. However, some users may still be wary of sharing personal data with Meta, especially given its history of privacy violations. The success of Muse will depend on whether users trust Meta with their personal information and whether the assistant's capabilities and security features are enough to overcome the company's past issues.