Meta is preparing to turn its smart glasses into a digital wallet, according to a reverse-engineering leak from Android Authority. Hours before the company’s Meta Connect keynote, evidence shows that Muse—the company’s new personal AI agent—is being tested as a replacement for Meta AI on smart glasses, complete with a native payment wallet called Meta Pay for Muse.
This report is based on a reverse-engineering leak, not an official announcement. While the track record of the source is strong, we are looking at code, not a finished product. Still, the implications of what is being tested are massive. We have already covered how three agents with shopping capabilities are defining this holiday season, but moving an agent from a phone in your pocket to a wearable on your face changes the game entirely.
The core issue here is the collapse of the traditional separation between observation and action. When you use an AI on your phone, there is a deliberate, physical barrier: you have to pull the device out, unlock it, and interact with a screen. With smart glasses, the agent is always on, always watching, and now, potentially always ready to spend your money. You are wearing a device that combines a live camera, biometric data streams, and a direct line to your bank account. That is a fundamentally different trust surface than anything we have dealt with before.
Meta is clearly aware of the optics. The leaked technical details point to a Secure VM architecture, which uses an isolated Linux environment and a Sentinel permission system to keep credentials outside the agent’s runtime. They are also promising per-transaction approval within the chat interface, ensuring that Muse never actually sees your card numbers. It is a robust attempt to build a digital fortress, but it faces a significant credibility deficit. Between the $18 billion multistate settlement from earlier this year regarding social media harms and a pending class-action lawsuit over privacy concerns with their glasses, Meta is not exactly starting from a position of neutral trust.
The market is already reacting to this tension. We have seen a clear split in how platforms are handling these new agentic capabilities. Amazon has blocked Muse from its shopping ecosystem, signaling a defensive stance against unauthorized agent commerce. Conversely, Shopify has leaned in, enabling Muse for agentic checkout across its stores. It is a fragmented landscape, and consumers are caught in the middle.
The data suggests that most people are not ready for this level of integration. According to recent research from Menlo Ventures, 76% of consumers cite privacy as a major concern, and only 20% are currently willing to grant an AI access to their financial accounts. Meta is trying to bridge this gap with technical safeguards, but for many, the privacy tax of an always-on, camera-equipped, payment-enabled wearable might be too high. Even the company’s plan to release camera-free smart glasses by late 2026 feels like a tacit admission that the current hardware is a hard sell for the privacy-conscious.
The technical details raise immediate, practical concerns. How exactly does the Sentinel permission system handle a scenario where the glasses are stolen? Can a software-based safeguard really protect a user when the hardware itself is compromised? And perhaps most importantly, how does Meta plan to reconcile the deep-seated privacy concerns of the majority with such an aggressive rollout of agentic commerce?
We have already seen Muse expand to macOS with computer use capabilities, showing that Meta is determined to put this agent everywhere. But putting it on your face, where it can see what you see and pay for what you want, is a different threshold. Tonight’s keynote will likely confirm the technical specs, but Meta faces a steep climb in proving that a software-based security layer is enough to overcome the inherent risks of wearable-based payments.