Muse, Meta’s New Personal AI Agent, Needs You to Trust It

Key Points:

• Meta has launched Muse, a personal AI agent available via a dedicated app, website, WhatsApp, and soon AI glasses, designed to automate digital tasks securely with built-in privacy features.
• Muse allows users to interact in natural language to perform tasks like sending emails, booking travel, and making purchases using Stripe’s Link payment infrastructure, which protects users’ financial information.
• The AI agent operates within a Secure VM architecture that isolates user activity to enhance security and privacy, with a future Confidential VM planned to give users full control over access keys and data encryption.
• Meta emphasizes responsible data handling through a system called Sentinel that monitors actions and requires user approval for sensitive operations, aiming to build trust despite the company’s historically challenged reputation on privacy.
• Muse is part of Meta’s broader AI strategy from its Superintelligence Labs, competing with other AI agents by focusing on security, privacy, and transparency, supported by extensive testing, external audits, and a public bug bounty program.