Will online sellers deliberately game an AI shopping bot's blind spots, like its bias for certain list positions, to boost sales?

This explores whether merchants will turn AI shopping agents' predictable quirks, especially their habit of favoring items in certain list positions, into a sales tactic. The corpus has no direct evidence of merchants doing this in the wild yet. It does show that the conditions for it are already in place, and that position may not even be the strongest lever. The starting fact is that position bias in agents is real and stubborn. The ACES experiments found that agents keep favoring particular slots even in plain-text interfaces with no visual layout, and even when told outright to ignore position Do AI shopping agents show persistent position bias in text interfaces?. A bias that survives an explicit instruction to ignore it is exactly what an optimizer would want to target.

There is a catch for anyone trying to exploit it. The same research found the bias is model-specific and shows up again, in a new form, after each model update. That makes it less like a fixed weakness and more like a search-engine ranking algorithm: you can game it, but you have to re-measure and re-tune every time the model changes. That points to a likely future that looks a lot like SEO, an industry built around testing and chasing each agent's current preferences. The 'agent attention economy' argument predicts this directly. As people hand goals to agents, services stop competing for human clicks and start competing to be chosen by agents, and new ranking and discovery infrastructure grows up around that competition Will agents compete for attention just like users do?.

The more surprising finding is that merchants may not need position at all. In a separate ACES test, a seller-side AI agent that only rewrote product descriptions, using what it knew about competitors, raised its product's market share by roughly 4 to 15 percentage points with five of the six buyer models tested. Price, rating, and position stayed the same Can AI seller agents shift buyer preferences through description edits?. Agents also react to cues far more sharply than people do. ABXLAB found that under the same price, rating, and nudge manipulations, agents fell into rule-like switching, while humans shifted only modestly Do shopping agents respond more strongly to price and rating cues?. And these biases aren't caused by attention or memory limits, as they are in humans. They appear to be patterns learned from training data and reward signals Where do LLM choice biases actually come from?. That makes them more mechanical, and so more predictable and easier to target, than human biases.

The exploitation also runs beyond product listings. Seller agents can work out a buyer's willingness to pay almost perfectly from conversation alone when the buyer agent is playing a shopping persona Can a seller infer what a buyer really wants to pay?. Buyer agents in negotiations often reveal their budgets unprompted and overpay Do LLM negotiation agents reliably follow budget and pricing rules?. So the weak point isn't one bias. It's a whole set of predictable behaviors on the buyer's side.

Whether this becomes systematic may depend less on the technology than on who pays for the agent. Azhar argues that an agent funded by merchant referral fees can't be loyal to both the merchant and the user, and that this incentive, not capability, will shape which agents get built Can an AI agent serve both merchant and user interests fairly?. Accenture's push for brands to build their own agents Will brands succeed by building their own AI shopping agents? points the same way. If the merchant owns the agent, it doesn't need to exploit a bias, because it can just set the ranking. The question to watch is less 'will merchants game position bias?' and more 'whose agent is doing the shopping?'