Mastercard opened developer early access for Agent Pay for Machines (AP4M) on October 20, 2026, releasing buyer- and seller-side APIs, a sandbox and developer toolkits. In a Cloudflare Connect demonstration, marketing agency Hype tested an autonomous agent making nine payments to two merchants within a single budget, without manual billing or account setup. AP4M uses HTTP 402 payment challenges, but differs from crypto-native implementations of the x402 protocol. Its agents issue digitally signed vouchers off-chain; merchants can batch and redeem them later instead of settling every request on-chain. The pilot combined agent wallet infrastructure from Coinbase Developer Platform and BVNK, settlement on Base, and Mastercard’s multi-currency fiat and stablecoin settlement platform. Circle separately expanded its managed payments service to target AI-agent traffic using USDC. Together, the developments show competing approaches to machine-to-machine payments, with different choices about settlement, funding and who manages digital assets.

Mastercard and Visa are advancing different approaches to AI-initiated business payments. At Cloudflare Connect, Mastercard demonstrated an agent making nine payments across multiple service providers under one authorized budget. It also opened early developer access to Agent Pay APIs, a sandbox and toolkits. Visa expanded its commercial agentic suite around a BILL and Billtrust pilot, alongside spend analytics and fraud tools. The distinction is how each approach constrains the agent. Visa’s pilot lets an agent pay invoices that buyers have already approved, using its Trusted Agent Protocol to navigate supplier-portal security without sharing underlying login credentials. Mastercard is testing controls for machine-to-machine transactions, including signed vouchers, agent verification and a score intended to estimate whether an incoming transaction was AI-initiated. Both efforts leave a consequential question for business payments: how responsibility should be assigned when an authorized agent follows its technical instructions but breaches a procurement condition.

AI agents can select products and initiate payments without a buyer present at checkout, forcing merchants and payment providers to verify delegated authority rather than a single approval. Google’s Agent Payments Protocol (AP2) uses signed mandates to record spending limits and connect instructions to a specific cart. Stripe and OpenAI’s Agentic Commerce Protocol and Shared Payment Tokens, alongside Mastercard’s Agentic Token framework, support checkout with restricted credentials. Visa’s Trusted Agent Protocol helps merchants identify agents, while Coinbase’s x402 enables stablecoin payments through web requests, including small API charges. These tools address different risks: signatures document permission, tokens limit payment use, and records can support disputes. But compliance with a spending cap cannot prove an agent chose what its user intended. Google transferred AP2 stewardship to the FIDO Alliance in April 2026, yet consumer deployment remains early. Merchants still face integration challenges, while liability and chargeback treatment for agent errors remain unsettled.

Coinbase demonstrated machine-to-machine AI payment workflows at Sibos 2026 in Miami Beach, highlighting the Model Context Protocol, Agent-to-Agent protocol and its x402 payment standard. The demonstration followed the August general availability of Amazon Bedrock AgentCore Payments, which incorporates Coinbase Developer Platform wallet infrastructure and x402 settlement. The AWS service supports Coinbase Developer Platform and Stripe Privy wallets for autonomous USDC microtransactions, while keeping developer credentials separate from AI agents and applying deterministic spending limits. The work connects a conference demonstration to infrastructure already available to developers. Coinbase and AWS had also enabled publishers and API endpoints using AWS CloudFront and AWS WAF to monetize autonomous agent traffic through x402. Coinbase Institute research argues that AI agents need sub-cent stablecoin payments, although that argument is not evidence of transaction volumes. The developments show how payment authorization, wallet access and settlement are being assembled for agent-initiated transactions, without establishing how widely the technology is used.