Artificial intelligence (AI) agents are set to take on a more prominent role in shopping for consumers as Google introduces its Agent Payments Protocol (AP2). This protocol functions as a neutral, open-source framework that allows both merchants and consumers, along with third-party platforms, to harness the benefits of AI. It supports various payment methods such as debit cards, credit cards, stablecoins, and real-time payments.

Concerns about granting AI agents complete autonomy in making purchases on behalf of users could be mitigated through the use of mandates—digital contracts that ensure AI follows a user’s instructions securely. For instance, if a user instructs an AI agent to purchase tickets for the upcoming baseball playoffs, they would draft a mandate specifying the desired price, timing, and other conditions. The initiator then signs another mandate that empowers the AI agent to carry out the transaction once all stipulated conditions are met.

A Robust Backing

The promise of AP2 would be strengthened with industry-wide adoption. Google has garnered significant backing from key players, including credit card companies like Mastercard and American Express, fintech giants such as PayPal and Alipay, crypto firms Coinbase and MetaMask, and established businesses like Etsy, Intuit, and Salesforce.

In total, over 60 companies have endorsed AP2, signifying a collaborative effort across the industry. Similar to Google’s recent blockchain initiative for financial institutions, its objective with AP2 is to provide an open, unbiased framework for all parties involved.

Building Trust Among Consumers

Though this support is noteworthy, questions remain about whether consumers will find value in using AI agents for shopping. While mandates can enhance trust and security, there are concerns over potential misuse by bad actors who exploit the system.

In AP2, Google has integrated safeguards to create an auditable trail, enabling reviewers to scrutinize fraudulent transactions if necessary. However, these measures might not be sufficient to fully convince consumers to relinquish control entirely to AI agents.

This is a fascinating topic and aligns with our recent research on agentic commerce,” said Don Apgar, Director of Merchant Payments at Javelin Strategy & Research. Several individuals quoted in this article question where liability lies when the agent operates outside its authorized scope: should it be attributed to the consumer, merchant, or card issuer? Additionally, we probe deeper into whether consumers can trust that an AI agent delivers the best deal for them, especially considering the potential for commissions from merchants.

Just look at Google’s search engine where companies pay for placement at the top of results even when their content might not be the most relevant,” he explained. Kudos to Google for establishing a framework that validates and securely operates AI agents but there remain substantial business and financial questions for consumers.”