Your AI agent has a wallet. Who answers for its decisions?

Your AI agent has a wallet. Who answers for its decisions?

That safeguard will not last, and it shouldn’t. An agent that needs your approval for every purchase is a faster checkout button, not an agent. Once approval moves from the transaction to the instruction, payments lose the assumption they were built on: that a human decides, at the moment of payment, that this is what they want. The industry is making rapid progress on how agents pay with existing payment methods. It has not yet solved what agents are allowed to decide, or who pays when they decide wrong. The companies that solve that problem will define the agentic economy. Authorized, delivered, and wrong Imagine asking your agent to plan a family vacation within a $5,000 budget. It books nonrefundable flights on the wrong weekend, a hotel 40 minutes from the beach, and accepts a counteroffer from the airline’s own agent that quietly strips out checked bags. Every payment was authorized. Every merchant delivered exactly what was ordered. The trip is still nothing like what you asked for. OpenAI’s terms leave delivery and returns with merchants and preserve existing card dispute protections. However, agents introduce a different kind of failure: a purchase that was properly authorized and fulfilled, but did not reflect the user’s instructions. Existing dispute mechanisms may offer remedies, but they do not clearly establish who should bear the loss when the failure lies in the agent’s judgment. Authorization to spend is not authorization to judge. Today’s infrastructure only knows how to verify the first. What the agentic stack is missing The industry is beginning to build verifiable mandates. Google’s Agent Payments Protocol, for example, uses cryptographically signed mandates to record what a user asked for and approved. That is necessary, but not sufficient. A signature can prove what was authorized. It cannot prove that the agent interpreted the user’s preferences correctly, acted in their interest, or was free of outside incentives. Closing that gap requires a broader idea of what the industry has started calling Know Your Agent. Today it means verifying whose agent is acting and what it was authorized to do. It must also mean verifying whether the agent exercised that authority as intended. That requires three layers. First, mandates that capture intent. In practice, agent permissions still center on price limits, timing and product parameters. A meaningful mandate also encodes intent: the user’s priorities, the trade-offs they accept and reject, and the conditions that require escalation, such as “nothing nonrefundable without asking me.” Relevant constraints must be verifiable across the transaction, by the merchant’s agent, the payment provider and any later arbiter, without unnecessarily exposing the user’s private instructions. A spending limit tells the system how much. A mandate tells it what for. Second, a liability framework. When an agent acts inside its spending authority but outside its mandate, the loss could land in many places: the user who gave an ambiguous instruction, the provider whose model misread it, the merchant whose agent extracted worse terms, the platform that ranked the options, or the payment provider that executed the transaction. No rule decides among them today, so every failure is likely to become a dispute. To earn consumers’ trust, the allocation should be settled in advance, for example by asking which party can prevent the harm at the lowest cost. Other industries have done this. The card industry’s EMV liability shift made the party that had not adopted chip technology bear certain counterfeit-fraud losses, creating a powerful incentive to upgrade. Product liability law allocates defects along the supply chain. Clear allocation also creates a market: risk that can be assigned can be priced and insured. Third, a duty of loyalty. OpenAI is not taking a cut of Wallet transactions at launch. But business models evolve. Once an agent’s platform earns money from merchants, through fees, paid placement or advertising, the agent faces a conflict of interest its user cannot see. Financial regulation addressed a similar conflict for brokers and advisers through disclosure and best-interest obligations. Consumer agents need enforceable standards for whose interests they serve, how paid influence is disclosed, and how compliance with the user’s mandate can be independently verified. Who will write the rules AI labs, payment companies and commerce platforms are writing the agentic protocols right now. Private standards are well suited to identity, credentials and payment flows, because every participant benefits from interoperability. They are structurally weak on liability, because every party at the table would prefer the loss to land somewhere else. I have worked on both sides of the rulebook, as chair of the FATF’s Risk Group and as a fintech executive. In both roles, the rules that worked were the ones that settled, in advance, who absorbs failure. Systems that answer that question scale. Systems that leave it open stall, or have the answer imposed by regulators after a visible failure, usually in terms written for the last crisis. The companies that build mandates, clear liability rules and loyalty into their agents now will not be slowed down by it. They will be the ones consumers trust with the wallet. Card payments became global infrastructure because people believed that when something went wrong, someone would make it right. Agents need the same promise, across consumers, AI platforms, merchants and financial institutions that must trust decisions none of them fully controls. Whoever makes that promise credible will do more than enable agentic payments. They will help write the rules of the new economy. The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.

Original Source

Read the full article at Fortune →

KhanList aggregates and links to publicly available news content. We do not host full articles from third-party sources. Always verify important information with original sources.