Two senators just asked the FTC to investigate whether Amazon’s Alexa for Shopping is quietly burying Made in USA products. If you sell on Amazon, this isn’t a policy story to skim past. It’s a preview of exactly how much your listing data now controls what AI shopping assistants tell customers, and most sellers still have no idea how exposed they are.
What actually happened
On September 17, 2026, Senators Tammy Baldwin (D-Wisconsin) and Rick Scott (R-Florida) sent a letter to FTC Chair Andrew Ferguson asking the agency to investigate Amazon’s Alexa for Shopping and Walmart’s Sparky chatbot over how they handle country-of-origin questions.
The letter leans on research from Columbia Law School’s Center for Law and Economy, which tested both assistants and found they can identify suspicious or false “Made in USA” claims, but do not reliably flag them as fraud or help shoppers filter for genuinely American-made alternatives. According to the senators’ letter, when researchers asked Alexa to find Made in USA products, it claimed it lacked access to that data. When asked for products made in China, it generated a list without hesitation. The letter also cites Alexa reportedly acknowledging internally that surfacing a Made in USA filter would “redirect significant sales away from their largest seller base.”
Amazon pushed back, stating it does not intentionally withhold country-of-origin information and that the details are available on individual product detail pages while the company continues improving Alexa’s accuracy. Walmart has not issued a public response. No formal FTC investigation has been announced as of this writing, and the senators’ letter is a request for scrutiny, not an enforcement action.
Why this matters even if you never mention “Made in USA”
It is tempting to read this as a niche compliance story for a narrow slice of domestic manufacturers. It isn’t. The real story here is what it exposes about how Alexa for Shopping, Rufus’s successor, actually decides what to tell a customer.
These assistants are not reading your bullet points the way a human shopper does. They are pulling from structured attributes, backend search terms, and whatever is explicitly stated in text fields. If a data point isn’t cleanly present, in the right field, the assistant treats it as unknown rather than inferring it, even when the answer is technically visible somewhere on your listing. That is precisely what the Columbia researchers found: the information wasn’t hidden out of malice, it was structurally invisible to the model in the way the assistant was querying it.
Country of origin happens to be the attribute making headlines this month. Next month it could be material composition, battery certification, or age grading. The underlying lesson applies to any factual claim you want an AI shopping assistant to surface correctly: if it lives only in a lifestyle image, a PDF spec sheet, or a vague adjective in your bullets, it may functionally not exist to Alexa for Shopping.
Where Alexa actually pulls this data from
Based on Amazon’s own statement and how the Product Detail Page schema works, country-of-origin and similar factual attributes come from a few specific places, in roughly this order of reliability:
- The dedicated “Country/Region of Origin” backend attribute in your listing data, when your category and template support it
- Explicit statements in the title, bullet points, or product description (“Made in the USA,” “Assembled in Vietnam”)
- Brand registry content such as A+ modules and Brand Story, if the claim is stated in text rather than baked into an image
- Structured Item Highlights and searchable attribute fields introduced alongside the 75-character title limit change earlier this year
What does not reliably feed the assistant: text embedded inside product images, PDF manuals, or Q&A answers left by other customers. If your only proof of origin is a small “Made in USA” badge baked into your main image, Alexa has no path to that claim at all.
A practical audit for your catalog this week
You don’t need to wait for the FTC to decide anything to tighten this up. A few concrete steps:
- Open your category’s flat file or listing template and confirm whether a Country of Origin attribute exists, and whether it’s actually populated for every SKU, not just your flagship ASIN
- State factual, verifiable claims (origin, material, certification) as plain text somewhere in your bullets or description, not only in imagery
- If you manufacture domestically, treat that as a searchable attribute worth stating explicitly rather than assuming shoppers will find it in an image
- Audit any claim you’re not fully confident you can document. The regulatory attention here cuts both ways: false claims are now more likely to get flagged, not less, as Amazon improves detection
- Check your A+ Content and Brand Story modules for the same gap. Premium A+ is now free for all brand-registered sellers, and factual claims placed there in text form are far more durable for AI retrieval than a matching graphic
None of this requires new software or a consultant to execute today. It requires someone actually opening the backend attributes tab category by category, which is exactly the kind of unglamorous listing work that keeps paying off as more shopping shifts to conversational search.
The bigger pattern worth watching
This is the second time in a matter of months that scrutiny of Amazon’s AI shopping layer has centered on a data completeness problem rather than a deliberate manipulation problem. As Alexa for Shopping becomes the default entry point for search-bar queries, the gap between “the information exists on my listing” and “the AI assistant can actually retrieve it” is becoming the thing that separates sellers who benefit from this shift and sellers who quietly lose visibility without understanding why.
Expect more attribute-specific attention like this over the next year, not less. Regulators, journalists, and competitors are all now capable of running the exact kind of query-based audit Columbia’s researchers ran on your own catalog. The sellers in the best position are the ones who treat structured data completeness as an ongoing compliance and merchandising task, not a one-time setup step from years ago.
Frequently asked questions
Does this mean I need to add a Made in USA claim to my listings?
No, only if it’s true and you can document it. The point isn’t to add the phrase, it’s to make sure whatever is factually true about your product’s origin, materials, or certifications is stated as text in a field the AI assistant can actually read, not just implied in an image.
Will the FTC investigation actually happen?
As of this writing, the senators have only requested an investigation. The FTC hasn’t announced one. Regardless of the outcome, the underlying data-completeness issue it surfaces is worth fixing on your own listings either way.
Where do I find the Country of Origin field in my listing?
It varies by category, but it’s typically in your product’s flat file template or in the Manage All Inventory edit view under a “Compliance” or “Country/Region of Origin” attribute. If your category template doesn’t expose it, you can usually still state it in your bullets or description.
Could stating false origin claims get my listing suppressed?
Yes, and increasingly so. Amazon’s own detection is explicitly called out in this story as capable of identifying suspicious claims. Unsupported or false claims are a suppression and account-health risk independent of this specific investigation.
If you want a second set of eyes on whether your catalog’s attribute data is actually complete enough for Alexa for Shopping to work in your favor, Depu Chakrabarti works with FBA and multi-channel sellers on exactly this kind of listing and PPC optimization work.
Sources:
EcomCrew: Senators Want an FTC Probe Into Amazon and Walmart’s “Made in USA” AI Answers
PYMNTS: Senators Demand Probe Into Amazon and Walmart AI Over Made in USA Claims