Zuckerberg: Muse AI Users Spend Over $200 Per Purchase, Sparking Payment Partner Surge

Mark Zuckerberg stated during Meta’s third-quarter earnings call that Meta Muse, the company’s AI agent, is being used for shopping by a significant portion of its user base. The agent now serves more than 10 million engaged users, and approximately one-quarter of them are conducting online purchases through the automated system.

What is the scale of this activity?

Zuckerberg disclosed that users are making purchases with an average value of over $200 per transaction. He noted that financial infrastructure provider Plaid observed activity levels through its connector that it had not seen with any other service. This volume was higher than Meta anticipated, requiring operational adjustments to handle the load.

How did payment systems respond?

Due to the unexpected surge in transaction volume, Meta worked directly with payment application Stripe to increase credit limits for users. This adjustment was necessary to accommodate the shopping demand generated by Muse users. The interaction highlights how AI-driven commerce can rapidly stress existing financial processing frameworks and require immediate vendor coordination.

What is the strategic goal?

Zuckerberg described this level of adoption as achieving “product-market fit” for Muse. He outlined a long-term vision where personal AI agents communicate directly with business AI agents to complete transactions without human intervention. Meta aims for Muse to eventually generate more revenue than its combined existing apps, including Facebook, Instagram, WhatsApp, Threads, and Messenger.

How does this relate to previous reporting?

As TechSec Feed reported on 2026-10-21, engineering improvements at Meta’s Ohio facility recently reduced daily GPU faults, boosting reliability for Muse model training runs. These infrastructure upgrades support the scaling of the agent’s capabilities. Meta also plans to integrate Muse into smart glasses launching next month, allowing users to access the agent without smartphones.

Questions to ask your suppliers

Security teams should ask vendors how their systems distinguish between human-initiated and AI-agent transactions. Determine if existing fraud detection models account for high-volume, automated purchasing patterns. Ask whether suppliers have implemented rate limiting or additional verification steps for non-human actors. Identify which internal teams, such as finance and legal, need visibility into these automated contract executions to manage liability and compliance risks effectively.

Background: AI agents

AI agents are software systems that perceive their environment, make decisions, and execute actions to achieve specific goals. They move beyond generating text to performing tasks like coding, scheduling, or data analysis. You must treat them as active participants in your network, not passive tools.

Read the full guide: AI Agents Examples: How Autonomous Software Changes Security

What to do and how to stay safe: Meta

• Review transaction logs for unusual spikes in volume or value that may indicate automated agent activity.
• Verify that payment gateways have appropriate controls for non-human initiated requests.
• Monitor vendor communications for updates on how they classify and secure AI-driven commerce.
• Ensure internal policies address liability for purchases made by third-party automated agents.

Step-by-step guide: AI Usage Policies at Work: How to Prevent Data Leakage

General security guidance from the TechSec Feed newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

How many users are shopping with Meta Muse?

More than 10 million engaged users are using Muse, with a quarter of them making purchases.

What is the average purchase amount?

The average purchase size is over $200, according to Mark Zuckerberg.

Did payment partners notice this activity?

Yes, Plaid reported unique activity levels, and Stripe raised credit limits to meet demand.