TLDR
Amazon has stripped the last product names from its order confirmation emails, replacing them with generic labels like '1 Electronics item'. At the same time, the company's robots.txt file bars at least ten AI crawlers from its product pages, cutting off third-party shopping agents while its own assistant, Rufus, operates freely inside the platform.
KEY TAKEAWAYS
The email that no longer says what you bought
Amazon's order confirmation email used to read like a tidy receipt: item names, product images, prices. Now it reads closer to a customs form filled in by someone who prefers not to commit. In early July 2026, the last customers still receiving full product details reported their order confirmation emails reverting to generic labels with no item names or photos.[1] A pair of noise-cancelling headphones becomes '1 Electronics item'. A children's book becomes '1 Books item'.
Amazon first stripped item details from order confirmation and shipping emails in mid-2020, leaving only the order number and price.[1] A portion of customers kept receiving richer emails in the years that followed, suggesting an A/B test stretched to an unusual length. The July 2026 rollout closed that gap entirely, making the stripped-back format universal.
What Google built, and what Amazon removed
The timing is pointed for Google. In September 2025, Gmail introduced its Purchases tab, a consolidated view designed to gather all purchase-related emails and package updates in one place. Gmail director Chris Doan pitched it as the way to stay on top of incoming deliveries and the best deals, timed for the holiday shopping season.[2]
Google launched the Gmail Purchases tab on 11 September 2025, consolidating purchase-related emails and package updates into a single streamlined interface, a feature that now receives almost nothing of substance from Amazon, the largest e-commerce platform most of its users shop on.[2] Third-party budgeting and receipt-parsing apps that scrape inbox data face the same problem: without item names, they cannot categorise purchases, generate spending reports or flag unusual charges.
Amazon's rationale, offered through Seller Central support, was framed in the language of user protection. An Amazon staff member answering sellers on the company's forums said Amazon "made updates to how customer information displays in Buyer-Seller Messages for privacy and security reasons", adding that the change was platform wide and "just the new normal we're all adjusting to".[4] Privacy is a defensible reason to reduce data in emails. Amazon's own internal systems are not subject to the same reduction.
The robots.txt wall and the crawlers it names
Email is only one of two channels Amazon has closed. The second operates at the infrastructure level, through a robots.txt file that instructs web crawlers which parts of a site they may visit. Amazon's robots.txt file disallows a wide range of AI and shopping-agent crawlers, including GPTBot, ClaudeBot, PerplexityBot, Google-Extended, GoogleAgent-Mariner, GoogleAgent-Shopping, Meta-ExternalAgent, Mistralai-User and Copilot, from crawling any pages.[3]
GPTBot serves OpenAI's indexing pipeline. ClaudeBot does the same for Anthropic. PerplexityBot powers one of the most widely used AI search products. GoogleAgent-Shopping sits behind Google's AI-assisted shopping features. Blocking them in aggregate means any AI assistant trying to answer a query like "what is the cheapest blender on Amazon" cannot retrieve live product data, prices or availability directly from Amazon's pages.
The robots.txt standard is technically voluntary. Crawlers are expected to honour it, and the major commercial operators do, but there is no enforcement mechanism short of IP blocking and legal action. The file states Amazon's preference plainly and at scale: these agents are not welcome.
Rufus and the asymmetry at the centre of the strategy
Amazon's own AI assistant, Rufus, operates inside the Amazon platform, drawing on Amazon's product catalogue, customer reviews and purchase history directly. Rufus is not a crawler subject to robots.txt constraints. Every restriction Amazon has placed on external data access leaves Rufus structurally unaffected, while competitors attempting to build comparable shopping intelligence from the outside are denied the inputs they need.
By severing the email data loop and the crawler data loop at the same time, Amazon closes the two main channels through which third parties could build an approximate view of its catalogue and its customers' behaviour.[3] An AI shopping assistant trained on real-time Amazon data would be a powerful product. Amazon has ensured that only Amazon can build one.
What this means for brands and developers
For merchants selling on Amazon and developers building tools that touch the e-commerce data ecosystem, the lesson is that platform-mediated data is always provisional. Structured email data, crawler access, API availability: all of it exists at the discretion of the platform and can be withdrawn without notice and without recourse.
The durable alternative is owned infrastructure. Self-hosted product pages with embedded schema markup remain accessible to customers and, where crawlers are permitted, to AI agents. Direct API feeds controlled by the merchant preserve continuity of data for internal analytics, AI features and compliance reporting. A brand that has built its data strategy around Amazon's email templates and its tolerance for third-party crawlers has built on ground that Amazon owns and can shift whenever it chooses.
The moves are individually defensible: reducing customer data in transactional emails is a legitimate privacy measure, and blocking AI crawlers is a legitimate exercise of site control. Together, they describe a company that understands precisely where value leaks out of its ecosystem and has decided to stop it. Whether that constitutes good platform stewardship or the consolidation of informational advantage around a proprietary AI product is a question regulators in several jurisdictions are increasingly equipped to ask.
SOURCES & CITATIONS
FREQUENTLY ASKED QUESTIONS
Why did Amazon strip product names from its order confirmation emails?
Which AI crawlers has Amazon blocked, and what does that mean?
Is Amazon's in-house AI assistant Rufus affected by these restrictions?
What can brands and developers do to protect their data access?
Nathan Cross writes about big technology companies and the economics of a generation coming up behind them. He is interested in scale, and in who pays for it.
