MoonPay Expands Crypto Payments Into ChatGPT as AI Commerce Gains Ground

Crypto payments are increasingly moving beyond wallets, exchanges, and traditional checkout pages as major technology and financial companies explore a new model: AI agents that can initiate and complete transactions on behalf of users.

The development is creating a new intersection between crypto payments and artificial intelligence. Instead of manually selecting a product, entering payment details, and confirming each step, users could increasingly tell an AI assistant what they want and allow it to handle the transaction within predefined limits.

MoonPay’s PayBox is one of the latest examples of this shift, connecting crypto and payment capabilities with AI assistants such as ChatGPT and Anthropic’s Claude.

MoonPay Brings Crypto Payments to ChatGPT and Claude

MoonPay has launched PayBox, a noncustodial payment vault designed to allow AI assistants to prepare and execute crypto transactions and online purchases.

Users can connect their PayBox vault to supported AI platforms and describe what they want in natural language. The AI assistant can then search for available options, prepare the transaction, and execute it according to the permissions established by the user.

The system supports several crypto-related actions, including purchasing digital assets with fiat, swapping tokens, moving assets between blockchains, and depositing funds into supported decentralized finance protocols.

PayBox also extends beyond crypto-native transactions. Depending on the available integrations, an AI assistant can use the system to book flights, make restaurant reservations, and purchase products from online retailers.

This effectively turns an AI assistant into a payment interface, allowing users to complete multiple financial and commercial steps through a single conversation.

PayBox Gives Users Control Over AI Spending

MoonPay has designed PayBox around two authorization models.

The “Always Ask” setting requires users to approve every transaction with a passkey. Each approval is limited to a single transaction, meaning the AI cannot reuse the authorization for another payment.

The “Autonomous” setting gives the AI assistant greater freedom to act within spending limits and other rules selected by the user. This removes the need for individual approval for every eligible transaction, while still restricting what the assistant can do.

Users must provide another passkey authorization if they want to change the permission model or transaction rules.

The approach reflects a broader challenge facing agentic payments: giving AI systems enough authority to be useful without giving them unrestricted access to a user’s money.

Crypto and Card Payments Move Into AI Agents

PayBox supports both crypto wallets and payment cards.

For wallet transactions, MoonPay uses multiparty computation technology to divide private keys across secure environments. This means neither MoonPay nor the connected AI system can independently reconstruct the complete key and authorize a transfer.

For card payments, PayBox uses Visa’s agentic commerce technology, allowing approved purchases to be completed without giving the AI access to the underlying card number.

The platform also supports Solana and several Ethereum Virtual Machine-compatible networks, including Ethereum, Base, Arbitrum, Polygon, Hyperliquid, Tempo, and Robinhood Chain.

MoonPay has also integrated the x402 payment standard, which is designed to enable services to accept payments initiated by AI agents.

The combination means an AI assistant could potentially execute several connected steps in one interaction, such as helping a user acquire a crypto asset, move it to another blockchain, and deposit it into a supported DeFi protocol, provided all actions comply with the user's predefined permissions.

Coinbase Pushes Crypto Payments Further Into AI

MoonPay is not alone in connecting crypto infrastructure with AI agents.

Coinbase has also been developing infrastructure that allows AI agents to interact with crypto wallets, make payments, and execute financial operations. Its Coinbase for Agents initiative enables developers to connect AI agents to Coinbase services and build applications capable of trading and making payments under defined controls.

The company has also developed AgentKit, which provides developers with tools for giving AI agents access to crypto wallets, while its x402 technology is designed to facilitate payments between AI agents and online services.

This is important because it moves crypto beyond the concept of an AI assistant simply recommending a transaction. The agent can potentially become the actor that actually executes it.

Coinbase is effectively working on both sides of the emerging ecosystem: enabling agents to make payments while also providing infrastructure for businesses to accept them.

Robinhood Opens Financial Services to AI Agents

Robinhood is pursuing a similar direction by opening parts of its financial infrastructure to AI agents.

The company has introduced tools that allow AI agents to interact with Robinhood services, including trading and purchases through its financial products, while maintaining security controls and limits around what agents are permitted to do.

Robinhood’s move illustrates another important development: AI agents could increasingly become the interface through which users access financial services.

Rather than opening an application, searching for an asset, selecting a transaction, and confirming the order manually, users could eventually tell an AI assistant what outcome they want and let it execute the required steps within predefined boundaries.

Visa and Mastercard Enter the Agentic Payments Race

The shift is not limited to crypto companies.

Traditional payment giants Visa and Mastercard are also developing infrastructure for AI-driven commerce, suggesting that agentic payments could become a major battleground across both traditional finance and crypto.

Visa has introduced Visa Intelligent Commerce, a system designed to allow AI agents to discover products, compare options, and make purchases on behalf of consumers.

The company is developing tokenization, authentication, and spending controls to help ensure that AI agents can transact securely. Visa has also been exploring ways for merchants to identify authorized agents and distinguish them from malicious automated activity.

The company is simultaneously expanding its work around stablecoins, AI, and tokenized payments, highlighting the growing overlap between traditional payment networks and blockchain-based infrastructure.

Mastercard is taking a similar approach with Agent Pay for Machines, which is designed to support payments initiated by AI agents and machines.

The system is being developed with a broad group of partners, including crypto companies such as Coinbase, OKX, and Tempo, alongside payments infrastructure providers such as Stripe.

Mastercard has also demonstrated AI-agent payments in collaboration with Banco Santander, showing that the technology is moving beyond experimental crypto applications and into regulated financial environments.

ChatGPT Is Becoming a Commerce Interface

The most significant development may be coming from the AI side itself.

OpenAI has been moving ChatGPT beyond information retrieval and product discovery toward actual commerce. Through Instant Checkout and its Agentic Commerce Protocol, developed with Stripe, users can purchase products directly within ChatGPT.

The initial rollout included Etsy merchants, with plans to extend the experience to a much larger number of Shopify merchants, including brands such as Glossier, SKIMS, Spanx, and Vuori.

This development is particularly relevant to crypto payments because it demonstrates that the AI interface is already becoming a place where transactions can begin and, increasingly, be completed.

MoonPay, Coinbase, Visa, and Mastercard are effectively working on the financial layer underneath that experience.

The result could be a future where the user interacts primarily with an AI assistant while the payment infrastructure operating behind the scenes becomes increasingly invisible.

Stablecoins Could Become the Payment Rail for AI Agents

Stablecoins could play a particularly important role in this emerging economy.

Unlike traditional payment systems that often depend on banking hours, geographic boundaries, and multiple intermediaries, blockchain-based assets can move continuously across networks.

That makes stablecoins potentially well suited to machine-to-machine and agent-to-agent payments, particularly when transactions are small, frequent, automated, or cross-border.

Visa has already highlighted the potential for AI agents to pay for services such as travel, data, and computing resources, while suggesting that stablecoins could provide programmable, always-available settlement infrastructure.

Mastercard is similarly exploring models that allow agentic transactions to use both traditional payment networks and stablecoins.

This suggests that the future may not necessarily involve crypto replacing cards. Instead, cards, stablecoins, and blockchain networks could operate as different payment rails underneath AI-driven commerce.

From Manual Checkout to Autonomous Commerce

The broader development is therefore bigger than simply adding crypto payments to ChatGPT.

The traditional payment process requires the consumer to perform almost every step: find the product, compare prices, enter payment information, authorize the purchase, and track the transaction.

Agentic commerce reverses that model.

The user provides an objective, such as booking a flight within a certain budget or purchasing a specific product, and the AI agent handles the research, selection, and payment within predefined rules.

Crypto infrastructure can make this model even more programmable by allowing assets to move directly between wallets, applications, and blockchain networks.

This creates an emerging financial stack in which AI becomes the interface, crypto and stablecoins can serve as programmable payment rails, and companies such as Coinbase, MoonPay, Visa, Mastercard, and Stripe provide the infrastructure connecting the two.

The Next Battle May Be Over the Infrastructure Behind Money

The most important aspect of these developments is not simply that ChatGPT or Claude can now help users make payments. The bigger shift is that the interface through which people interact with money is beginning to change.

If users can tell an AI agent to book a flight, purchase a product, acquire a digital asset, or move funds between blockchains, the payment application itself could become less important to the user. The real competition may move underneath the interface, toward the infrastructure that gives AI agents secure and controlled access to money.

This is where crypto and stablecoins could gain an advantage. Their value is not only that they provide another way to pay, but that they represent programmable assets that can move across blockchain networks around the clock.

The race, therefore, may not be about building the best crypto wallet or payment app. It could increasingly be about who builds the infrastructure that allows AI agents to access, move, and settle money safely and autonomously.