AI shopping assistants are rapidly evolving, though they currently pause before completing online payments, requiring human intervention at checkout. Nike’s release of Caitlin Clark's first signature shoe, the Caitlin 1 in "Caitlin Blue," exemplifies this trend. The shoe nearly sold out within two hours, and the article’s author used an AI agent to track down a pair. The agent efficiently searched multiple retailers, confirmed sizing preferences, and placed the shoes in an online cart at Dick’s Sporting Goods, but stopped short of finalizing payment, highlighting the current technological and regulatory limits around autonomous transactions.

This cautious approach stems from infrastructural challenges around secure payments by AI agents. New protocols such as Google’s Agent Payments Protocol and standards developed by Stripe and OpenAI aim to establish transparent, digitally signed permissions that safeguard consumer interests. Retailers who adopt these protocols gain visibility and competitive edges in the growing agentic commerce market, predicted by McKinsey to reach $1 trillion in U.S. retail revenue by 2030. Despite these advances, widespread consumer trust remains limited, with only about a quarter of online adults willing to let AI handle routine purchases.

Beyond consumer shopping, AI agents are also initiating machine-to-machine microtransactions using stablecoins like USDC. These digital dollars, supported by blockchain, enable near-instantaneous, low-cost payments for services such as data feeds or computational resources—transactions too small and too numerous for traditional card networks. Industry players, including Circle, Visa, Mastercard, and Google, are collaborating through initiatives like the x402 Foundation and Nanopayments framework to scale and standardize these machine payments, indicating a significant shift in the financial infrastructure underpinning AI commerce.

However, this new paradigm introduces complex security, fraud, and liability issues. AI agents could be manipulated to make unauthorized purchases or funnel money to illicit actors, and stablecoins’ rapid settlement complicates traditional dispute mechanisms. Legal frameworks are just beginning to address responsibility when agents err—whether liability lies with the user, AI developers, payment providers, or merchants. Policies around agent permissions, real-time fraud screening, and consumer protections must evolve alongside technology to ensure AI-powered commerce grows safely and transparently, especially as these agents increasingly handle critical financial decisions.