Why do shoppers want final say before an AI agent actually spends their money?

This explores why people want a final sign-off before an AI agent spends their money. The demand is real and specific. In RTB House's survey, pre-purchase approval was the safeguard consumers wanted most from shopping agents, and 44% of baby boomers preferred it Do consumers want to approve AI shopping purchases before they happen?. The obvious explanation is that payment is hard to undo, and actions that are irreversible and visible tend to trigger stronger demands for oversight. The rest of the corpus points to subtler reasons. Consumers will hand an agent the browsing, but they are not yet sure the agent is on their side.

Start with the trust gap. A Harris Poll for Visa found that 72% of US adults use AI assistants but only 23% trust AI to execute payments. When a familiar payments brand was attached, 61% said they trusted that company to handle agent-initiated transactions Does naming a payments brand increase trust in AI transactions?. Accenture found the same split from another angle. 74% said they would trust an agent with purchases, yet 56% would still specify brands themselves Do consumers actually want to delegate shopping decisions to AI?. Taken together, people are comfortable letting agents search and compare. They want to keep the commitment step, either for themselves or for an institution they already trust.

That instinct looks well founded once you see who might be paying the agent. Hedges warns that agents funded by advertising have built-in reasons to favor paid placements, however well they are engineered Do advertising-funded shopping agents favor paid product placements?. Azhar argues that an agent earning referral fees from merchants cannot be fully loyal to both the merchant and the user Can an AI agent serve both merchant and user interests fairly?. Jaipuria adds that placement payments could be folded into platform take rates, which would make them invisible to the user Could agent advertising hide inside platform take rates?. If the bias can't be seen during the search, the approval screen may be the only place a person can catch it. The agent can also give its user away. In one experiment, a seller-side agent inferred a buyer's willingness to pay almost perfectly from the shopping conversation alone Can a seller infer what a buyer really wants to pay?. A final check lets the person see what price their agent actually accepted.

There is a deeper reason too. Bilgihan and colleagues argue that agents which actually make purchases don't just carry out fixed preferences. They help shape those preferences through repeated back-and-forth Do AI shopping agents reshape what consumers actually want?. Seen that way, approval is a chance to ask whether this is really what you want, not only whether the agent got it right.

The infrastructure side is reaching a similar conclusion from engineering principles. The IMF argues that the agent's probabilistic decision-making should be kept separate from deterministic authorization and settlement, because payment systems need predictability and legal enforceability Should AI agents make payment decisions independently?. McKinsey proposes cryptographically signed mandates to replace the human signals that fraud systems used to rely on Does agent-initiated commerce break existing payment infrastructure?. What the reader might not expect is how easily the checkpoint disappears. Ayeshha argues that the bigger risk isn't hackers tricking the agent through prompt injection but a single "allow always" checkbox. In a reported case, Meta's Muse completed a sale and shared a user's address without checking back Do broad consent settings pose bigger risks than prompt injection attacks?. What consumers are asking for is a protection that one convenience setting can quietly switch off.