Shoppers say they trust AI agents to shop for them — so why do many still want to approve the purchase before it happens?

This explores why many shoppers want a final sign-off step before an AI agent spends their money, and what that wish reveals beyond simple caution. The headline number is smaller than you might expect. In RTB House's survey, about 35% of consumers named pre-purchase approval as the safeguard they want most, and the figure rises to 44% among baby boomers Do consumers want to approve AI shopping purchases before they happen?. The usual explanation is that people want more oversight when an action can't be undone and other people can see it. Buying something fits both, because it moves money and commits you to a seller. A recommendation you can ignore. A completed purchase you have to reverse.

The desire for approval also sits alongside a strong willingness to delegate, which looks contradictory at first. In Accenture's survey of 25,590 people, 74% said they would trust an AI agent with purchases. Yet 56% would still name the brands, and 37% of loyal customers wouldn't switch at all Do consumers actually want to delegate shopping decisions to AI?. So people are happy to hand over the tedious searching while keeping the final judgment. Approval rights are how they draw that line. Research on high-autonomy shopping assistants points the same way: people accept them when the benefits feel concrete, especially when scarcity makes speed valuable, not because they feel in control What drives shoppers to accept autonomous AI assistants?. Approval is a hedge that lets people take the benefit without fully giving up control.

There is also a less obvious reason: whose side is the agent on? MIT work argues that agents built by platforms inherit the seller's goals by default, while agents owned by consumers can be set up around the buyer's goals Does agent ownership determine whose interests get optimized?. Bob Hedges warns that agents funded by advertising have a built-in pull toward paid placements, however well they are engineered Do advertising-funded shopping agents favor paid product placements?. Leakage is a related risk. An agent that personalizes well can reveal what you're willing to pay just through how it behaves. Sellers have recovered a buyer's target price almost exactly from conversation alone Can a seller infer what a buyer really wants to pay?, and prompt-level privacy instructions can't stop this, because the leak comes from the personalization itself Can better prompts stop preference leakage in AI shopping?. Read this way, the approval step is a checkpoint against an agent whose loyalties and discretion you can't verify.

The twist is that approval rights are fragile in practice. Ayeshha argues that the bigger danger isn't prompt injection. It's the "allow always" checkbox, which quietly overrides per-purchase approval. In one 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?. The law doesn't fill the gap yet. EU consumer law was designed around human thinking and decision-making, so its tests and disclosure rules fit poorly when an agent is the buyer Can EU consumer law protect AI agents as algorithmic buyers?. Owens and Law argue that agents should owe the same fiduciary duties as human brokers before markets lock in Should AI shopping agents face fiduciary duties like human brokers?. Until something like that exists, the approval click is doing work that law and agent design should be doing.

The most interesting idea in the collection is that the approval moment may be where your preferences actually take shape. Bilgihan and colleagues argue that agents which carry out purchases end the consumer's role as the sole author of their own wants. Preferences get built through back-and-forth loops between the agent's learning and the shopper's trust Do AI shopping agents reshape what consumers actually want?. If that's right, approval rights protect more than your wallet. They protect the last point where the choice is clearly yours. That also explains why platforms may push for full autonomy before it is in consumers' interest: a formal model shows platforms gain conversion revenue from agentic search earlier than consumers rationally should switch Why do consumers delay adopting agentic search?.