Skip to main content
Back to News
Important
Operations

White House proposes an AI border to catch tariff evasion

August 19, 2026
6 min
Summarize with AI
A

CEO at Nova Analytics

LinkedIn

Antoine founded Nova Analytics to empower Amazon sellers with enterprise-grade analytics. He specializes in data architecture and building scalable solutions for e-commerce businesses.

Nova surfaces every Amazon fee, refund, and margin shift in your live P&L, across 23 marketplaces. Check the SKU-level breakdown

What happened

A White House trade report argues that illegal transshipment of China-linked goods through third countries has become a systematic way to avoid US tariffs, and proposes an AI-enabled "Detective Border" to catch it. The system would combine shipment data, routing history, ownership links, product classification, production-capacity indicators and anomaly detection to flag high-risk entries for Customs and Border Protection (FreightWaves, August 18, 2026).

The report names more than 40 countries as presenting elevated transshipment risk, from Mexico, Canada, the EU, India, Japan, South Korea and Taiwan through to Vietnam, Malaysia, Thailand, Indonesia, Brazil and Turkey. It points to Executive Order 14411, signed June 3, 2026, as the enforcement framework, with tougher bonding, ownership-disclosure and good-standing requirements on importers.

The scale being claimed

The report draws on five separate analyses and puts potential annual transshipment exposure somewhere between roughly $40 billion and $303 billion, noting the estimates use different methods and are not additive. A Commerce Department analysis inside it estimated about $67 billion of US-bound goods transshipped through Mexico, India and Vietnam in 2025, implying roughly $28 billion in lost tariff revenue.

Ranges that wide usually mean the enforcement response arrives before the measurement is settled. That is the part importers should plan around.

Why this lands on sellers who did nothing wrong

Most Amazon sellers do not choose a transshipment route. They choose a supplier, and the supplier chooses the route. If your goods were legitimately manufactured in Vietnam or Malaysia, you are still moving through a corridor the model has been trained to treat as suspicious.

An anomaly-detection system does not distinguish between a small importer with thin documentation and a deliberate origin launderer. It flags entries. The cost of a flag is a hold, and the cost of a hold in September is a stockout in November.

The report also explicitly targets limited processing that creates the appearance of a new country of origin without meeting the substantial transformation test. If your supplier moved final assembly to a third country after 2018 and you never checked the origin analysis, that is worth reading before Q4 inventory ships.

What to do before peak shipping

  1. Ask every supplier for the origin basis in writing. Where the components come from, where the substantial transformation happens, and which HTS classification they are claiming.
  2. Check whether your production country changed recently. A supplier that moved from China to a neighbouring country after 2018 is exactly the pattern the model is built to surface.
  3. Build clearance delay into the Q4 timeline. Add buffer weeks rather than assuming last year's transit time.
  4. Put duty into landed cost per unit, per shipment. Duty is not a fixed percentage any more, and a blended annual rate hides which SKUs are actually underwater.
  5. Know which SKUs you cannot afford to have held. Rank by daily contribution margin, then protect those first with earlier shipping or a second source.

How Nova helps

  • Amazon P&L - duty and freight land in cost per unit, so a tariff change shows up as a margin change on the SKUs it actually touched.
  • FBA Inventory - days of cover per SKU, which is the number that tells you how long a customs hold can last before it costs you sales.
  • Custom Breakdowns - group SKUs by supplier or origin country to see your exposure in one view.