Amazon AI scraping project creates unapproved list of small brands

The intersection of artificial intelligence and e-commerce reached a new point of friction this week, as independent retailers discovered their proprietary product data was being collected for Amazon's secret efforts. This program has been identified by industry observers and affected sellers as: project starfishincludes automated scraping of independent business websites to generate AI-powered listings within the Amazon Shopping app.

Independent brands such as Bobo Design Studio and Hitchcock Paper have reported that the e-commerce giant is currently testing a beta program to acquire products from small business websites without the owner's consent. These retailers discovered that Amazon was using AI to mine information from the site and create lists where sellers did not opt ​​in or could not opt ​​out. This discovery has sparked a debate about intellectual property, the accuracy of generative AI in retail, and the boundaries of platform dominance in the digital age.

How Project Starfish works and agent discovery

At the core of this development is a technical framework designed to expand Amazon's reach into inventory that Amazon does not store or ship directly. Amazon is currently scraping thousands of independent websites through Project Starfish to create AI-generated lists within its own applications. These products are often labeled to users as “Buy it yourself” or “Buy direct from other stores.”

The system works by deploying an AI agent that scans the open web for products that are not currently available on the Amazon store. Once a product is identified, the system “reskins” the independent company’s products and funnels them into the Amazon-branded checkout process. This marks a major shift towards the adoption of agentic AI, where autonomous software components perform complex workflows, including purchases, on behalf of users.

Technical documentation and seller reports indicate that the program creates a layer of abstraction between consumers and actual sellers. Depending on the product, customers can shop on the site with the help of AI or go directly to the site to complete the transaction. However, the automated nature of this process introduces significant technical differences that impact brand reputation.

AI illusions and data integrity challenges

One of the main technical failures reported by affected retailers has to do with inaccuracies in AI-generated content. Amazon's AI agents have been shown to “hallucinate” product details, displaying factually incorrect but authoritative information. These illusions include showing old prices or selling items that are no longer in stock.

For small businesses, seeing incorrect pricing or out-of-stock products can quickly disrupt operations. When an AI-generated listing offers a product at a price that is no longer acceptable to the seller, the seller is forced to choose between incurring financial loss and disappointing new customers. Additionally, according to a.pdf, the program uses AI images that are not owned by the original creator. According to a.pdf, this practice breaks the chain of trust for customers and disrupts the pricing and fulfillment models established by indie shops.

AI psychedelics are not unique to Amazon's retail experiment, but the risks become even greater when linked to financial transactions. Other platforms face similar challenges, such as hallucinations in TV search, where inaccurate responses are mitigated by verification servers. In the context of Project Starfish, it does not appear that the merchant whose data is used has access to such a validation layer.

Proxy email issues and customer relations

A major technical hurdle for merchants taking orders through the program is the lack of direct customer data. Merchants are left with “ghost orders” from anonymous proxy emails, making it impossible for them to provide support or build real customer relationships.

In traditional e-commerce, the relationship between buyers and sellers is the basis for handling returns, exchanges, and shipping updates. By hiding a customer's true identity, Amazon maintains control over the data, but independent sellers are responsible for fulfillment and potential complaints. Some industry analysts see this data collection as an attempt to gather information to inform Amazon's own merchandising and targeted advertising.

This change in data ownership is consistent with a broader industry trend in which first-party data collection is prioritized to circumvent privacy restrictions. By positioning itself as an intermediary for off-platform products, Amazon effectively captures signals of shopping intent that would otherwise remain in its own domain.

The institutional hypocrisy and scraping debate

Project Starfish’s findings highlighted a perceived hypocrisy in platform policies. Amazon currently blocks AI companies from scraping its own data as well as collecting intellectual property from independent sellers.

Amazon has been proactive about securing its data perimeter. The company expanded its bot limits on November 24, 2025, blocking OpenAI-related crawlers from accessing Amazon.com. This defensive posture suggests that while Amazon views its own product data as proprietary property that must be protected from competitors, it views independent retailer data as a public resource for its own AI training and list generation.

This tension is also reflected in the legal world. For example, xAI recently filed a lawsuit against the state of California over a law that forces AI companies to disclose training secrets. The conflict between the need for transparency in AI training and the protection of trade secrets remains a central point of contention for all major technology platforms.

Strategic context within Amazon's AI ecosystem

Project Starfish does not exist in a vacuum. This is part of a larger push towards autonomous shopping. On November 18, 2025, Amazon announced a comprehensive update to its AI shopping feature, saying that the number of products available for purchase through “Buy for Me” has increased from 65,000 at launch to more than 500,000.

The company has also integrated the Rufus AI assistant, which manages price tracking and automated purchasing features. These tools are designed to reduce friction with consumers, often by bypassing traditional brand-consumer interactions. Amazon describes these features as helping customers “discover and evaluate products,” but the retailers involved claim these features have been “hijacked” without their permission.

Impact on the industry and the future of retail media

For the marketing community, these developments herald a new era of “unlicensed” retail media. Traditionally, retailers chose to list on Amazon and paid for that privilege through commissions and advertising fees. With Project Starfish, that option no longer exists.

The growth of Amazon DSP and the introduction of tools like Amazon Marketing Cloud demonstrate the company's focus on unifying the shopping experience across the web. However, if this integration occurs without the consent of product owners, it raises questions about the long-term viability of the independent web.

Some large retailers have opted for formal partnerships, such as Macy's, which integrated Amazon's retail advertising technology in late 2025. This allows retailers to maintain control of their customer relationships and data. The contrast between these formal agreements and “agency scraping” of small and medium-sized businesses highlights the deep gulf in how platforms exercise their power.

The arrival of Project Starfish and Shop Other Stores Directly has fundamentally changed the way digital advertising and product discovery works. For PPC professionals and digital marketers, Amazon's traditional “walled garden” has expanded its walls to include its own website.

According to PPC Land, the rise of agent AI means a brand’s SEO and product data is no longer just the preserve of the human eye or standard search engines. These are now being actively captured by shopping agents who reskin and resell that content. This creates a “leaky bucket” of traffic where potential customers may find a brand’s products via Amazon’s AI listings rather than visiting the brand’s site directly.

Marketers now need to consider “Agent Optimization.” Being ranked on Google is no longer enough. When Amazon agents scrape your site, they must ensure that the information they retrieve is accurate. But as the Hitchcock Paper and Bobo Design Studio report shows, even accurate data can be distorted by AI “illusions” once it enters Amazon's ecosystem. This loss of control over brand presentation, pricing, and customer data represents one of the most significant challenges for independent e-commerce in the last decade.

timeline

• August 2024: Amazon DSP introduces modeled attribution for non-Amazon conversions and lays the foundation for tracking the entire customer journey across the web.
• January 9, 2025: Amazon launches Retail Ad Service, allowing other retailers to use Amazon's advertising technology.
• November 4, 2025: Amazon has filed a legal claim against Perplexity AI, demanding that it block its AI agents from making purchases on Amazon while also expanding its own scraping agents.
• November 11, 2025: Amazon introduced free AI prompts to sponsored ad campaigns to automate shopper engagement.
• November 18, 2025: Amazon has updated its AI shopping feature to reveal that its “Buy for Me” program now includes more than 500,000 products, many from sources other than Amazon.
• November 24, 2025: Amazon extends robots.txt limits to block crawlers from Meta, Google, and OpenAI.
• December 2025: An independent seller went viral on Instagram and LinkedIn after it was revealed that his website was being scraped without permission via Project Starfish.

summary

• who: Amazon and its own AI agents are the main actors, impacting thousands of independent retailers and small business owners such as Bobo Design Studio and Hitchcock Paper.
• what: Project Starfish and the beta program known as “Shop Other Stores Directly.” It uses AI to scrape independent websites, create unapproved product listings, and facilitate purchases through Amazon's “Buy for Me” agent interface.
• when: The program saw significant expansion in late 2025, with key adoption metrics released in November 2025 and widespread seller discovery in December 2025.
• where: This activity takes place on the open web (scraping) and within the Amazon Shopping app (listings and checkouts) and primarily targets independent brands based in the United States.
• why: Amazon aims to expand its product assortment, collect competitive product data, and capture first-party shopper signals, but this comes at the expense of seller consent and data accuracy.