Story
September 10, 2026
Meta’s Muse Asks Users to Trade Trust for AI Convenience
Meta argues that Muse can turn an AI assistant into a useful, tightly controlled partner for everyday life; skeptics see a product that demands extraordinary access from a company still burdened by privacy failures. The launch makes convenience and trust inseparable.
Meta’s push into personal AI arrives after years in which its privacy record has repeatedly shadowed its product ambitions. The company has faced regulatory penalties, lawsuits and the enduring fallout from scandals including Cambridge Analytica—history that makes a more autonomous assistant a harder sell than another chatbot.1
On Tuesday, Meta launched Muse in the U.S. for adults, positioning it as an agent that can take on the digital chores people usually handle themselves: sending emails, booking trips, completing forms, shopping and monitoring connected services. Zuckerberg called it “the personal agent that understands your goals and works 24/7 to get things done for you.”
2
The commercial pitch is broad. Muse is available on iOS, Android, the web and WhatsApp, with support for Meta’s AI glasses planned. A free tier is meant to cover most users, while $20 and $100 monthly plans offer heavier usage. Meta says the product is a step toward “personal superintelligence,” with Zuckerberg promising free access up to 100 million tokens a week.3
Meta’s central argument is that Muse is not simply being given the keys. It operates in a dedicated cloud virtual machine with a visible browser; a separate system, Sentinel, decides whether actions can proceed, be blocked or require approval. The company says passwords and payment details remain out of the agent’s view, and users can limit permissions by task, service or time period.4
Yet the tension is built into the design. Muse becomes more useful as it connects to email, calendars, payments, social accounts and third-party services—and Meta will use interactions to train models unless users opt out. The company says it strips critical identifying information before training, but the burden remains on consumers to decide whether its safeguards outweigh the data they are being asked to share.5
For Meta, Muse is a bid to bring agents beyond developers and into daily life. For users, it is a practical question with an unusually loaded premise: can the company that knows so much about them be trusted to act on their behalf?