People say they'd trust an AI agent to shop for them — but would they actually let it buy without checking first?

This explores whether saying you trust a brand turns into letting an AI agent actually buy things for you. The short answer is that the corpus can't settle it. Almost all the evidence here comes from surveys of stated trust, and the few behavioral studies suggest that real delegation depends on other things. The surveys are striking on their own terms. Accenture found 74% of consumers say they'd trust an AI agent with purchases, but 56% would still name the brands themselves, and over a third of loyal customers wouldn't switch at all (Do consumers actually want to delegate shopping decisions to AI?). So people trust the agent to do the errand, not to choose what they buy. RTB House's 'super-trusters', who would let AI buy with no return protection, are a minority, not the norm (How much do shoppers trust AI shopping assistants?).

The most interesting pattern is that brand trust seems to act as a stand-in for trust in the AI, not as a sign of it. In Visa's Harris Poll, only 23% of US adults trusted AI to make payments, but 61% trusted Visa to handle payments started by an agent (Does naming a payments brand increase trust in AI transactions?). People aren't saying 'I trust AI now.' They're saying 'I'll go along with it if a familiar company is standing behind the transaction.' Accenture's finding that consumers prefer agents built by retailers points the same way (Will brands succeed by building their own AI shopping agents?). Its authors also admit that these are survey preferences, not evidence that the agents work at scale.

When researchers watch what people actually do, different factors come into play. In a hands-on study, trust dropped sharply on tasks that couldn't be undone and that other people would see, like sending an email. That happened even when people rated the agent's work as fine, and high-stakes tasks that could be fixed afterwards caused no such drop (What makes people distrust AI agents they delegate to?). A purchase is exactly the kind of action that can't easily be undone and is visible to others, so this may be where stated brand trust breaks down. On the adoption side, a set of five studies that included real ad click-through data found that perceived benefits, especially under scarcity, predicted uptake of high-autonomy shopping assistants (What drives shoppers to accept autonomous AI assistants?). That's a sign that the payoff people expect matters more than their general attitude. There may also be a sensible reason to wait: one formal model shows platforms gain from agentic search before it pays off for consumers (Why do consumers delay adopting agentic search?).

Two further findings complicate the question itself. First, the trust people state at the start is unstable. People initially avoid an AI partner once they learn it's an AI, but that bias reverses after repeated interactions where they can see the results (Does revealing AI identity help or hurt user trust?). Delegation may therefore follow experienced outcomes more than any trust measured up front, and trust signals can be shallow anyway: users rate answers higher simply because they carry more citations, relevant or not (Do users trust citations more when there are simply more of them?). Second, delegation may wear down the brand loyalty that was supposed to make it feel safe. LLM shopping agents shift their choices far more sharply than humans in response to price, rating and nudge cues (Do shopping agents respond more strongly to price and rating cues?). Agents that make purchases may also reshape what consumers prefer through repeated conversations (Do AI shopping agents reshape what consumers actually want?). So the brand someone trusted when they handed over the task may not be the brand the agent ends up choosing.

The honest gap is that no study here tracks the same people from their stated brand trust to whether they actually delegate a purchase. The best working hypothesis from the corpus is this: brand trust gets people to agree to delegate in principle. Whether they keep doing it depends on whether purchases can be reversed, what benefits they see, and the outcomes they experience.