# Mastercard Expands AI Agent Controls for Automated Payments

By Simon Yoon

Canonical URL: https://www.tokenpost.com/news/technology/30334
Published: 2026-10-12T13:10:37.000Z
Updated: 2026-10-12T13:10:37.000Z
Section: Technology

> The company is testing a transaction score in the United States while liability for mistaken purchases within approved limits remains unresolved.

Mastercard is expanding controls for AI agents that make payments on behalf of consumers and businesses, addressing the risks of mistaken purchases as automated commerce develops.

The company announced Agent Pay on April 29, 2025, using tokenized credentials that let verified AI agents purchase goods and services for users. Consumers can set what an agent may buy, while transactions are registered across the payment chain.

Mastercard launched Agent Pay for Machines on June 10, 2026. The service is designed for programmatic, high-frequency payments between software systems, including transactions worth fractions of a cent. It supports cards, accounts and stablecoins, with tools for agent credentialing, permissions, spending limits, transaction execution and settlement.

More than 30 companies were listed as early participants or supporters, including Coinbase, OKX, Stripe, Cloudflare and Solana Foundation.

On Sept. 30, Mastercard announced a probability score intended to estimate whether a transaction was initiated by an AI agent. The service is being tested in the United States. Mastercard has not disclosed the score’s accuracy, testing results, participating issuers or a date for general availability.

The company’s Agent Pay Trust Framework combines identity, intent, controls, execution and intelligence. “AI agents will make commerce more intuitive, efficient and personal — but only if people and businesses can trust the systems acting on their behalf,” Ann Johnson, Mastercard’s executive vice president, Security Solutions, said.

The risk extends beyond unauthorized fraud. An agent could remain within an approved budget but misunderstand a user’s instructions, such as booking the wrong flight or purchasing the wrong product. Mastercard’s public announcements do not establish whether the consumer, issuer, merchant, AI developer or Mastercard would absorb the loss in such a case.

In a Monday interview, Ed McLaughlin, Mastercard’s president and chief technology officer, discussed human approval, spending limits and controls enforced across the payment network. “Part of it is bringing humans back into the loop,” he said.

No confirmed case of a Mastercard-connected AI agent going on a rogue spending spree has been identified. Mastercard also has not publicly set a general liability standard for erroneous purchases made within an agent’s authorized limits.

Jorn Lambert, Mastercard’s chief product officer, said, “Machine payments can make it possible for services to be bought and sold among agents at fundamentally different scales than payments today.”

Agent Pay for Machines is aimed at consumer purchases and machine-to-machine payments, including digital services and computing resources. Mastercard reported more than 3.7 billion accounts and more than 175 billion transactions facilitated across its global network in 2025.