AI shopping agents fooled by fakes

AI agents powering product search and commerce continue to grapple with rampant hallucinations, fabricating citations and product links that severely undermine consumer trust. This issue is not theoretical; as researchers Minghao Luo and Liang Chen demonstrate, even a single fake web page can manipulate AI shopping recommenders—including those used by giants like Target, Walmart, Google, and OpenAI—causing them to promote fabricated brands as authoritative. Luo highlights the alarming ease of this manipulation, noting, “You only write one page out of 10” to significantly sway AI outputs, which fuels a deepening trust crisis across retail and data-driven industries.

The vulnerability of AI shopping bots to misinformation is starkly illustrated by real-world underground operations that churn out fake online reviews and counterfeit brand pages. Luo references a Chinese TV report revealing that a fake brand can ascend to the top recommendation of mainstream AI systems within hours, underscoring how minimal manipulation—such as replacing the top three search results—can increase the success rate of fooling AI recommenders to a staggering 73.8%. This active exploitation highlights the urgent need to address hallucination-induced misinformation that threatens the integrity of AI-driven commerce.

By early 2026, AI agents have distinctly bifurcated the commerce journey into discovery and transaction phases, prioritizing trust through evidence-based recommendations rather than traditional influencer persuasion. As Karen Webster observes, this shift underscores a fundamental change in how consumers engage with products, with AI platforms like ChatGPT or Gemini becoming the new starting points for discovery before directing shoppers to retailer sites such as Walmart.com or Kroger. This evolving dynamic positions brands to leverage AI and social commerce platforms to cultivate direct consumer relationships during discovery, while retailers maintain their critical role in transaction execution and basket-building.

The functional split between discovery and transaction is reshaping the retail value chain, compelling both brands and retailers to adapt their strategies to thrive in an AI-driven ecosystem. Mondelez highlights that while the majority of sales volume still flows through retailers, those who adeptly integrate AI-powered discovery tools with seamless transaction capabilities are poised to capture greater market share. As Mondelez notes, "There are brilliant retailers who are adapting to this extremely well and they're likely going to be the ones that continue to grow, share and do well," signaling that success hinges on harmonizing AI’s dual roles rather than viewing discovery and transaction as isolated functions.

By mid-2026, consumer enthusiasm for AI shopping tools is evident, with nearly half of American shoppers trying AI-driven options primarily to find the best prices and enjoy personalized curation that traditional platforms like Amazon may not fully offer. For instance, 22% of surveyed consumers appreciated AI’s ability to customize recommendations, highlighting a growing appreciation for AI’s tailored shopping experiences despite its novelty.

Despite high satisfaction rates—94% of American users reported positive purchase experiences—trust remains a formidable barrier, as over three-quarters express concerns about data privacy and security, particularly regarding AI chatbots’ handling of personal information. This skepticism is echoed internationally; UK and Australian consumers similarly hesitate to grant AI full autonomy in purchasing decisions, with fears ranging from unauthorized transactions to potential breaches of bank security.

Consumers continue to value the tactile and exploratory nature of traditional shopping, expressing reservations about fully replacing these experiences with AI, which they perceive as lacking the subjective judgment and serendipity inherent in human shopping. As one consumer noted, AI ‘lacks a soul’ and cannot replicate the nuanced style advice or surprise elements that make shopping enjoyable, underscoring a persistent preference for control and discovery over automated recommendations.

Transparency and trust in AI shopping ecosystems are critical, with consumers demanding clear identification of sponsored content and equitable AI preferences free from brand pay-to-play influence. Trust is evenly distributed among tech giants like Google and Apple, payment providers such as PayPal, and marketplaces like Amazon, reflecting a complex landscape where consumers weigh security assurances—83% expect AI payment security to match or exceed current standards—against concerns about monopolistic control potentially skewing search results and limiting open access.

Shopify is spearheading the agentic AI commerce revolution with its Universal Commerce Protocol and AI-powered platform that seamlessly integrates real-time data infrastructure into retail workflows. By redefining product discovery and elevating data quality as a competitive edge, Shopify accelerates the retail AI arms race against giants like Google, while also driving a transformation in market research and workforce skills, as highlighted by VP of Data Nell Thomas. This strategic push positions Shopify not just as a commerce enabler but as a pivotal force reshaping how AI agents interact with retail ecosystems in 2026.

Amazon’s agentic AI initiatives, exemplified by its Rufus assistant, aim to replicate and enhance the personalized in-store shopping experience through targeted questioning, tailored recommendations, and detailed product comparisons. Beyond customer-facing tools, Amazon is embedding generative AI across retail operations—from inventory management to sizing recommendations—supported by foundational AI technologies like chips, Sagemaker, and Bedrock. CEO Andy Jassy’s optimism about AI’s transformative potential is tempered by caution over the rapid pace of change, underscoring the high-stakes nature of the ongoing retail AI arms race.

The collaborative efforts of AI leaders such as OpenAI, Google, and Anthropic with e-commerce infrastructure giants like Shopify and Stripe are laying the groundwork for AI-native checkout systems and real-time inventory updates, heralding a new era of autonomous agentic commerce. These AI agents are envisioned to proactively anticipate consumer needs—handling tasks like weekly grocery runs or preparing gear for upcoming trips—thus fundamentally altering retail workflows and consumer interactions. However, despite this promise, current AI shopping assistants like Walmart’s Sparky remain only marginally better than traditional search bars, highlighting a significant gap between AI’s potential and present capabilities.

Agentic AI commerce is not only reshaping retail media networks by intensifying competition over purchase channels but also revolutionizing advertising through the integration of user experience memory with advertiser data such as inventory and CRM systems. This fusion enables more personalized and effective product search and ad experiences, shifting away from static sponsored content to dynamic, data-driven AI-powered advertising that better serves both consumers and advertisers. By focusing on critical customer moments—moments of despair and delight—companies like Perplexity are redefining retail interactions, signaling a broader transformation in how AI-driven workflows enhance user satisfaction and engagement.