Agentic payments: how AI agents will prove a human approved their purchases

Agentic payments are transactions an AI agent initiates and completes for a person within limits that person has set. The concept addresses three questions that traditional card systems were not designed to answer: whether the user authorized the agent, whether the request reflects the user's true intent, and who is accountable for errors or fraud. Several competing systems are emerging, including OpenAI and Stripe's Agentic Commerce Protocol, Google's AP2 framework built around signed mandates, Mastercard's Agent Pay, and Visa's Intelligent Commerce. Most of these narrow card credentials through tokens tied to a specific agent, merchant, or amount, while keeping a record of what the user approved. For merchants selling online, the shift means they may increasingly see that an order came from an agent, and they will need to handle those orders within these new frameworks.