MoonPay has released an open-source wallet standard designed to let AI agents hold funds and execute transactions across blockchains, addressing a key gap in how autonomous software interacts with crypto systems.

According to Monday’s announcement, the standard introduces a shared way for AI agents to access and use wallets across tools and blockchains, replacing fragmented setups where each system manages its own keys and balances. It allows agents to operate from a single pool of funds rather than across multiple disconnected accounts.

“AI agents can employ standard building blocks, such as APIs, to communicate with other agents and humans, receive and send money, and access and interact with the internet,” according to researchers at MIT Sloan.

MoonPay said recent efforts to enable machine-driven payments focus on transaction rails but do not address how wallets and keys are managed.

The new system stores private keys in an encrypted local vault and signs transactions in an isolated process, keeping keys out of the AI agent’s runtime. It also includes policy controls that let users set spending limits and restrictions before transactions are approved.

The standard is open source and modular, with components covering storage, signing, policy controls and chain support, and is available through developer platforms including GitHub, npm and PyPI.

Founded in 2019, MoonPay is a financial technology company that provides infrastructure for businesses and consumers to move funds between fiat and digital assets, offering services such as on- and off-ramps, trading and crypto payments across global markets.

The company said more than a dozen companies contributed to the new specification, including PayPal, OKX and Circle, alongside several blockchain foundations and infrastructure providers.

Related: MoonPay launches enterprise stablecoin suite with M0, taps ex-Paxos leaders

Companies expand tools for AI-driven crypto transactions

Crypto companies are increasingly building infrastructure to support AI agents as economic actors.

In a separate announcement on Monday, BitGo, a digital asset custody and infrastructure company, said it had launched a Model Context Protocol (MCP) server that allows AI-driven tools to access its developer platform using natural language, enabling agents to navigate wallet functions, transaction flows and staking systems.

The integration connects BitGo’s infrastructure to AI-native development environments, allowing tools such as ChatGPT and code editors to retrieve documentation, API references and product information directly within workflows.

The move reflects a broader push to integrate crypto services into AI systems, as companies experiment with ways for software to interact with financial infrastructure without relying on traditional user interfaces.

Other efforts have focused on enabling machine-driven payments, including Coinbase’s x402 protocol, which enables stablecoin transfers over HTTP for APIs, apps and AI agents, as well as tools launched last week by Visa and Stripe-backed Tempo that allow AI systems to initiate payments and execute transactions programmatically.

Magazine: Are DeFi devs liable for the illegal activity of others on their platforms?

This detailed match analysis covers key moments, player performances, and tactical insights.