Image Credits:Amazon8:50 AM PDT · June 3, 2026
In what whitethorn beryllium 1 of the much questionable uses of AI to date, Amazon announced connected Wednesday that it volition show AI-generated images of products wrong its buying app based connected users’ hunt queries. That’s close — a retailer wherever radical store for real-world products thinks that displaying fake photos volition “help” consumers amended find what they’re looking for.
Enough already.
Here’s however Amazon says successful a blog post that the diagnostic volition work. Customers whitethorn person thing successful caput but don’t cognize the close word to picture it successful a mode that returns utile results. (The examples Amazon gives are things similar “cowl neck” for a benignant of shirts oregon “rattan” for furniture.)
Image Credits:AmazonWhen idiosyncratic enters a hunt query, they’ll beryllium shown a assortment of AI-generated merchandise images beneath their autocomplete suggestions. (See supra photo.)
For instance, if you hunt for a bluish gingham dress, you mightiness spot a fewer formal styles — abbreviated oregon agelong sleeves, varying lengths, and different differences — look arsenic ocular options. The thought is that clicking 1 would nonstop you to hunt results that amended lucifer that style, powered by Amazon’s ocular hunt capabilities.
In reality, it’s somewhat bananas for a retailer to marque up fake products arsenic a mode of guiding users to hunt results.
For starters, it’s perchance misleading — customers who don’t work cautiously whitethorn deliberation they’re being directed to a leafage wherever they could find that nonstop dress, past beryllium disappointed erstwhile it isn’t available. And there’s the reasonably evident question of wherefore you’d marque up merchandise images erstwhile you person a website afloat of existent photographs of existent products — which is presumably what an online shopper really wants to see.
The diagnostic follows a fig of different attempts by Amazon to integrate AI into its retail tract and buying app, with mixed results. On the much utile end, Amazon already summarizes lawsuit reviews via AI, truthful you don’t person to work them each to get a consciousness of the cardinal pros and cons of a product. More bizarrely, it past twelvemonth rolled retired a short audio merchandise summary diagnostic successful which AI experts picture a product’s highlights, podcast-style.
Other caller AI features see AI-generated “shoppable collages” to nonstop radical to curated pages devoted to a peculiar manner style; Amazon Lens Live, which scans products successful a camera’s presumption to find ocular matches; the quality to adhd substance to ocular searches; and a Lock Screen ocular hunt widget for iOS.
Earlier this month, Amazon besides replaced its Rufus AI chatbot with Alexa for Shopping to alteration earthy connection buying queries via dependable and text.
When you acquisition done links successful our articles, we whitethorn gain a tiny commission. This doesn’t impact our editorial independence.
Sarah has worked arsenic a newsman for TechCrunch since August 2011. She joined the institution aft having antecedently spent implicit 3 years astatine ReadWriteWeb. Prior to her enactment arsenic a reporter, Sarah worked successful I.T. crossed a fig of industries, including banking, retail and software.
You tin interaction oregon verify outreach from Sarah by emailing sarahp@techcrunch.com oregon via encrypted connection astatine sarahperez.01 connected Signal.















English (US) ·