Card Networks Push AI Checkout While Shoppers Hesitate

Mastercard and Visa are accelerating efforts to integrate artificial intelligence agents directly into the checkout process, aiming to capture a role in commerce where bots shop on behalf of humans. Mastercard recently launched a feature allowing users to issue virtual cards to AI agents with specific spending limits and merchant restrictions. This move is part of a broader race by payment giants to establish themselves as the default infrastructure for autonomous transactions, a strategic shift that could reshape how digital money moves for decades.

Despite this corporate momentum, consumer sentiment remains cautious. Recent surveys indicate that while shoppers are eager to use AI for finding deals and comparing prices, the majority are uncomfortable delegating the actual act of purchasing. The industry is currently navigating a gap between the technical capability of autonomous agents and the psychological comfort level of the average customer who still wants to hit the final "buy" button.

Infrastructure race defines future standards

Industry observers describe the current developments as a land grab for controlling the rules of agentic commerce. Phil Bruno, chief strategy officer at ACI Worldwide, noted that if these card networks set the standards for how AI agents handle payments, they can lock in their dominance for the next generation of shopping. Visa is pursuing a similar path through its partnership with Alchemy and its own Intelligent Commerce product, which is currently in deployment. Meta has also entered the space with Muse, a tool that navigates checkout processes but still requires user approval before the transaction completes.

The strategic goal is to make the payment layer invisible and automatic, much like how contactless payments became standard. By embedding their protocols into the AI agents themselves, Mastercard and Visa aim to ensure that any future automated purchase flows through their networks. This positions them not just as payment processors, but as the gatekeepers of trust and identity in a bot-driven economy, a role that carries significant long-term financial and regulatory weight.

Consumers prioritize control over automation

Data from ACI Worldwide reveals a clear preference for human oversight among shoppers in the US and UK. Only seven percent of respondents said they would allow an AI assistant to make purchases without their approval, even when predefined conditions were met. More than half of surveyed consumers expressed discomfort with letting algorithms spend their money. Instead, the most valued features were price-drop alerts and cross-retailer comparisons, indicating that users see AI as a research tool rather than a spending agent.

This hesitation is not merely about trust in technology, but about the final act of commitment. For many, the checkout process is the culmination of research and decision-making, a moment where they assert control over their budget. Allowing a bot to complete this step removes that final checkpoint, creating a psychological barrier that marketing efforts alone may not overcome quickly. The gap between what companies are building and what customers are ready to use remains significant.

Liability questions remain unresolved

A major catch in these new systems is the ambiguity of liability when things go wrong. Mastercard has introduced a digital record called Verifiable Intent, which logs exactly what an agent was authorized to buy and by whom. The company claims this creates a paper trail that can help resolve disputes, ensuring that the same protections available in traditional card purchases apply to AI-driven transactions. However, the system relies on the cardholder’s instructions being clear and accurate.

When asked who is responsible if an agent makes a fraudulent or unauthorized purchase outside of those instructions, Mastercard pointed back to the Verifiable Intent record rather than specifying a direct party. This leaves a potential gap in accountability that could fall on the consumer or the merchant, rather than the payment network. Until it is clear who pays for mistakes made by autonomous agents, many shoppers will likely continue to retain the final click, ensuring that human oversight remains the standard for the foreseeable future.