AI agents can now make secure card payments with user-defined limits using Mastercard’s authorization system.
Artificial intelligence agents are steadily moving from simple assistants to tools capable of executing real financial actions. Payment networks and developers are now working to close that gap. A new collaboration between Mastercard and Crossmint aims to bring structured oversight into this emerging space.
AI Agents Gain Mastercard Payment Access via Crossmint Integration
Crossmint’s Lobster.cash is set to partner with Mastercard Agent Pay and Verifiable Intent, introducing a new way for AI agents to carry out payments. In a release, the setup is said to allow agents operating on open platforms to use a consumer’s existing Mastercard without exposing card details.
Moreover, the system combines card network infrastructure with a standardized method for confirming user approval. It extends trusted payment rails into fast-expanding agent ecosystems.
According to the statement, open-source platform OpenClaw will be the first to support the setup. Users can assign tasks to agents, including purchases, while maintaining clear boundaries. Each transaction passes through Mastercard’s network, meaning issuer controls and authentication processes remain active.
At the core of the system sits Verifiable Intent. Built in collaboration with Google, it creates a cryptographic record of user approval. That record shows who authorized a transaction, what rules were set, and whether those rules were followed.
As such, issuers and merchants can independently verify each payment without relying on platform-specific data.
Security concerns have slowed the adoption of agent-based payments in the past. However, this model removes the need for users to share card credentials directly with AI systems.
Instead, permissions act as the control layer, while Mastercard’s infrastructure handles authentication and monitoring. This structure keeps transactions traceable and reduces the chances of misuse.
Developers Gain Controlled Payment Logic for AI Agents
Lobster.cash acts as a shared payment layer across multiple agent platforms. It already connects with tools such as Claude Code, Devin, Hermes, and Zo Computer. Through it, users can program exactly how their agents behave financially. That includes where funds can be spent, how often transactions can occur, and which payment methods are allowed.
According to Crossmint co-founder Alfonso Gómez-Jordana Mañas, the goal is to keep things familiar for users.
“Mastercard Agent Pay is one of the most trusted payment infrastructures designed for agentic commerce in the world. Bringing it to lobster.cash means agent users don’t need a new wallet or a new card.”
Alfonzo said
From Mastercard’s side, the integration builds on existing infrastructure and is not a separate system. Chief Digital Officer Pablo Fourez stated that developers can create new applications while still operating within the same guardrails used in traditional payments.
Meanwhile, adoption of the solution is already underway. Mastercard Agent Pay is used by banks such as Santander, Commonwealth Bank of Australia, DBS, and UOB. Adding lobster.cash brings the same system into open platforms where developers and AI agents are rapidly increasing.