6 hours ago
108
Meta launched Muse on September 8, 2026, and the tech world’s reaction landed somewhere between fascination and mild existential dread. The personal AI agent can browse the web, fill out forms, and complete purchases on a user’s behalf, all inside a sandboxed virtual environment called the Muse Secure VM. It is, essentially, a digital errand runner that never sleeps and never forgets your shoe size.
PayPal is positioning itself to be part of that future. The company is working toward enabling its customers to use Muse for shopping and checkout globally, a move that would put PayPal inside one of the most ambitious AI commerce experiments currently running.
How Muse actually works
Think of Muse as a very capable assistant who has been given controlled access to your digital life. It connects to services like email and calendars, reads context, and acts on your behalf. The key guardrail: it requires explicit user approval before completing purchases or other sensitive actions.
At launch, Muse routes financial transactions through Stripe’s Link service. The mechanism uses one-time-use virtual cards, which adds a layer of buyer protection that traditional checkout flows don’t offer. Muse will also support Shopify’s Shop Pay in the near future, broadening where and how it can complete transactions.
The Muse Secure VM is the architectural choice that makes all of this possible without triggering immediate privacy alarms. Rather than giving the agent unrestricted access to a user’s machine, Meta runs Muse’s browsing and purchasing activity inside an isolated environment. User consent gates every sensitive step.
The subscription is tiered, running from $20 per month for standard access up to $100 monthly for expanded functionality. Muse launched exclusively in the U.S. and is restricted to users 18 and older, available through dedicated apps and the web.
Early traction, early friction
The market responded quickly. Muse climbed to the top of the Apple US App Store rankings shortly after launch, a signal that consumer curiosity about agentic AI is well ahead of what most industry observers had expected.
Amazon disagreed with the whole premise. Effective September 21, 2026, Amazon blocked Muse from shopping on its platform, citing violations of its privacy and security conditions. That is a significant wall to hit, given that Amazon represents a substantial share of U.S. e-commerce volume.
What agentic commerce means for the payment stack
Agentic commerce is the idea that software, rather than a human, initiates and completes a purchase. Traditional payment flows are designed around human decision points. A person sees a product, adds it to a cart, reviews the total, and clicks confirm. Fraud detection systems, chargeback rules, and consumer protection laws all assume a human is somewhere in that chain. When an AI agent is the one clicking confirm, after receiving a delegated instruction from a user who may have issued it hours earlier, the liability and trust questions get complicated quickly.
Stripe’s one-time virtual cards are a practical answer to part of that problem. They limit exposure on any single transaction and make it easier to dispute charges without compromising a user’s underlying payment credentials.
PayPal has been developing its own AI commerce capabilities separately, including in-chat payment features, suggesting it is approaching the agentic shift from multiple angles rather than betting everything on a single partnership.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) ·