UFLPA Entity List Reaches 190 as DHS Adds Three BGI-Affiliated Companies

The Analysis

The notice says this. The Department of Homeland Security published an updated UFLPA Entity List as of 29 October, and the roster now stands at 190 entities. Three companies were added: BGI Genomics, MGI Tech and Shenzhen BGI Technology. All three were listed under section 2(d)(2)(B)(ii), and two of them were also listed under 2(d)(2)(B)(v). Goods made wholly or in part by a listed entity carry a rebuttable presumption of forced labour and are barred from entry unless the importer proves otherwise, and the presumption reaches inputs several tiers upstream. The notice sets out a removal procedure for listed entities. It does not state what evidence an importer must produce to rebut.

I read notices like this one for a living, and my first instinct is always to separate the thing that changed from the thing that did not. Very little changed here. The statutory mechanism is the same mechanism it was before 29 October. The presumption, its rebuttable character, its reach into the supply base, all of that predates this document. What changed is three names, and the number at the bottom of the page. Everything else in the coverage is commentary on a mechanism that has been standing since the list was created.

That distinction matters because the two categories carry different clocks. A change in the mechanism can be commented on, litigated, delayed, phased. A change in the list is operative on publication, applies to merchandise as of the stated date, and offers the importer no procedural pause at all. If you plan your year as though the second kind behaves like the first, you will be late every time, and you will be late by exactly the amount of notice you assumed you were owed.

So let me do what I always do with a document of this kind: establish precisely what it says, then establish precisely what it does not say, and then be clear about which of the two is going to cost you money. Strictly speaking, the second question is the more expensive one, and it is the one the coverage skipped.

Take the arithmetic first, because the arithmetic is where the framing goes wrong. Three additions to a list of 190. That is 1.58 percent growth. If you are an importer and you read that as a 1.58 percent change in your risk, you have made an error of category, and it is the most common error in compliance reading. A list is a stock. Your exposure is a flow, and the flow is set by how many lines in your bill of materials, and in your suppliers' bills of materials, terminate at one of those names. Three names can be a rounding error for one company and an existential event for another, and the list total tells you nothing about which company you are. The only number that matters is not on the page, and it is the one you have to produce yourself: how many purchase orders, at how many tiers, trace back to these three entities.

Here is the structural point the coverage missed, and it is worth more than any of the numbers. Compare this document to the other customs instrument that landed in the same fortnight. The CBP proposal on low-value shipments was published on 8 October with a comment deadline of 7 December. That is a rulemaking, and rulemaking has a shape: a proposal, a comment window, a final rule, an effective date, and in between all of that, a period in which the regulated party can say something and be heard. This entity listing has none of that shape.

It is an administrative act. There is no comment window, no transition period written into the text, no phase-in by tariff line or by contract vintage. It appears, and it applies. Practitioners who are used to the rhythm of rulemaking consistently misjudge these listings, because the tools they have learned to use, the comment letter and the request for delay, do not exist here. The only procedural device in the document runs the other way: it is a removal procedure for the entity, not a relief procedure for the importer.

Next, the two clauses, because two of the three names sit on both and that is not decoration. The notice lists all three under 2(d)(2)(B)(ii) and adds two of them under 2(d)(2)(B)(v) as well. What the notice does not do is explain the facts that satisfy each clause. It does not tell you which conduct put a company on the first limb and which additional finding put two of them on the second. I am not going to invent that reasoning, and anyone who supplies it to you without a citation is guessing. What can be said with confidence is the consequence for the party carrying the burden: where two limbs apply, the importer faces two questions rather than one, and the answers are cumulative rather than alternative. You do not get to pick the easier question. You have to close both.

Now the sentence that does the real work in this document, and the one I would underline twice. The presumption reaches inputs several tiers upstream. Several. That is the word. Not two, not three, not a defined number that a compliance officer can map onto a supplier tier diagram. Several. In a legal instrument, an undefined quantitative term is not a small drafting gap, it is where the entire cost of compliance is decided, and here the cost is decided by an adjective. Sit with what that means for a moment. The importer bears a burden of proof, the burden extends upstream by an unspecified distance, and the standard of proof is not stated. Those three facts together do not describe a rule. They describe a range of possible rules, and the importer has to comply with the widest reading of the range, because the widest reading is the one that stops the container.

There is a second undefined term working with the first, and the combination is what makes this document hard to price. Goods made wholly or in part. In part has no threshold attached. It does not say a material part, or a substantial part, or a part exceeding some share of value. A trace input, an intermediate that constitutes a fraction of a percent of the finished article, is inside the phrase as written. Multiply an unbounded reach by an unbounded materiality threshold and you get a compliance obligation whose boundary cannot be drawn from the text. That is not a criticism of the drafters so much as an observation about what the importer is actually being asked to do: prove a negative, about an unbounded set, without a stated test. Anyone who tells you this is a manageable administrative exercise has not tried to prove a negative across a multi-tier supply base.

There is a third gap, and it is about identity rather than scope. The list operates on legal persons. The supply chain operates on corporate groups. The headline of this update describes the three as affiliated, and they plainly are, but what the document does is name three legal entities, and the presumption attaches to goods made by a listed entity. That distinction will settle most of the arguments had at a customs desk over the next month. A distributor buying from an affiliate that is not named is not, by that fact alone, inside the presumption. A purchase order that never mentions the group is not, by that fact alone, outside it. Identity is a documentary question, answered with a certificate of incorporation and a manufacturer's declaration, not with an inference from a brand name. Importers who reason from brands get this wrong in both directions, and both directions cost: one stops clean cargo, the other releases cargo into a proceeding that will outlast the shipment.

Then the date. The list is expressed as an update as of 29 October. Read that carefully, because as of is doing something. It anchors the operative moment to 29 October, not to the day the notice appeared in front of you. For a shipment that entered before that date and is still in the pipeline, and for a shipment that entered after it but before anyone in your organisation read the notice, there is a gap, and the gap is where disputes live. I am not going to overstate this. The notice does not say anything about goods in transit, and the absence of a transit carve-out is itself information: it means you cannot assume one. Practitioners who assume a transition window because a rulemaking would have had one are importing an assumption from the wrong instrument.

Now the asymmetry, which is the point I would put at the top of a board memo. The presumption is rebuttable. That is the law, and it is genuinely a limit on the provision, not a formality. But rebuttable on paper and rebuttable in practice are two different things, and this notice widens the distance between them. The entity that has been listed gets a procedure: a route to petition for removal, with the government obliged to respond to it in some fashion. The importer gets a presumption and a burden and no stated evidentiary standard. Consider who is better placed to prove the facts.

The listed company knows its own factory, its own inputs, its own payroll. The importer knows a purchase order number and a supplier's assurances. The party with the weaker access to the evidence carries the burden of producing it, on a per-shipment timeline, at a border. That misalignment is not a loophole anyone can exploit. It is a structural feature, and it means the practical rebuttal rate will be lower than the legal standard implies, for reasons that have nothing to do with the merits of any individual shipment.

Layered by who actually gets hit, this is not an apparel story, and reading it through an apparel lens will mislead you. These three entities sit in genomics: sequencing instruments, reagents, lab consumables, and the service contracts that keep them running. The importers exposed are laboratories, hospital groups, research distributors, contract research organisations, and the equipment vendors who ship a system and then supply it for a decade. Two consequences follow from that profile and neither is obvious from the headline. The first is that a reagent is a manufactured good with a short shelf life and a recurring order pattern, which means the exposure is not a one-off shipment but a monthly flow, and a stop on the flow is an operational stop, not a delay.

The second is that the reach into several tiers upstream runs straight through electronics and instrument assemblies, where components are sourced from contract manufacturers whose own supplier lists are frequently not disclosed to their customers. The importer at the top of that chain is being asked to certify a fact four steps down, held by a company that considers the fact commercially sensitive.

Which brings me to what a rebuttal would have to look like, since the notice declines to say. Three families of standard exist in this corner of law and practice, and they are not interchangeable. One asks about origin: prove the goods were not made by the entity, a documentary exercise with a defined answer. One asks about the transaction: prove this shipment did not pass through the listed facility, narrower and cheaper but silent on the general flow. One asks about the system: prove the supply chain is audited and free of the practice, durable and expensive.

A notice that states no standard leaves the importer guessing which of the three the officer at the port will accept, and the rational response is to prepare for the most expensive one, because a file is cheap against a stopped container and the standard can always be raised after the fact. That is an inefficient allocation of compliance spend. It is also the allocation the text produces, and pretending otherwise is how compliance budgets come to be built on optimism.

A boundary has to be stated, though, and here it is. Services are not merchandise. A laboratory in the United States that buys sequencing as a service from an affiliated provider is not making an entry, and the customs presumption does not reach that transaction by its own force. That transaction may raise other questions, under procurement rules or funding conditions, but those are different questions with different decision-makers and different remedies, and I have watched compliance teams tie themselves in knots by merging the two. If your exposure is a service contract, this notice is background, not the operative instrument. If your exposure is a physical good with a listed entity somewhere in its chain, this notice is the operative instrument, and the rest is commentary.

To the counterfactual, because a rule of this kind is only worth its weight if you can say what would change the conclusion. Suppose the three names had not been added. Your exposure would not be zero. It would be 187 other entities, the same presumption, the same unbounded upstream reach, the same silence on the evidentiary standard. The list is a stock and the mechanism is the constant, which is why I resist the framing that this is a new risk. It is an old risk with three new names and a fresh reminder that the bottleneck in these cases is almost never the list. It is documentation depth. An importer with a tier-two map and a legal manufacturer named on every component can rebut; an importer without one cannot, regardless of how clean the goods actually are. The list is the occasion for the loss. The documentation is the cause.

There is a respectable argument on the other side and I want to give it its due. The presumption is rebuttable, the entity has a removal route, and the history of these lists includes removals. A lawyer advising a client could reasonably say that the correct posture is patience: comply with the process, build the file, petition where the facts support it. I do not disagree with any of that as a matter of law. My disagreement is about timescale. The removal route operates on the entity's clock, measured in petitions and reviews. The importer's problem operates on the shipment's clock, measured in a container sitting at a terminal.

A remedy that arrives after the flow has stopped has not remedied the flow. Derek Xu puts the operational version of this better than I can: at the desk, the hold arrives without a reason code, and by the time the reason surfaces, the file you need is the one that was never assembled. That is the same asymmetry, seen from the other end of the building, and it is why I keep insisting that a rebuttable presumption is a legal fact about the goods and an evidentiary fact about the importer's paperwork, and only one of those is within the importer's control.

So what does the document actually require of a careful importer, stated as narrowly as the text allows. It requires that you know, for each article you import, whether any listed entity made it wholly or in part, at any tier the government might later assert. It requires that you be able to demonstrate that answer with evidence you can produce inside the life of a detention. And it requires that you accept, going in, that the standard you will be measured against is not written down. Three requirements, and only the second one is something you can build. The first is an inquiry, the third is a risk position, and the middle one is a file. Most importers spend their effort on the inquiry and the position and forget that the file is the only thing that gets opened at the border.

There is one piece of information I would pay for and cannot get from this document, and it is the cadence. How often the list is revised, and whether revisions cluster around particular policy moments. A stated cadence would let an importer schedule a review. The absence of one means the review has to be continuous, which in practice means it gets done once a year, or after a loss. Three names in a single update tells you the flow exists. It does not tell you the rate, and the rate is what a planning calendar needs. Read every one of these notices as a sample from an unknown distribution, because that is what it is.

I will end where I began, on the distinction between the mechanism and the list. The mechanism has been in place long enough that no importer can claim surprise about its existence. The list changes, quietly, on dates that appear in a Federal Register notice and nowhere in your planning calendar. The exposure that follows is not proportional to the number of names. It is proportional to the depth of your own visibility, and that depth is a decision you make years before any particular name appears. Read the notice for what it is: three names, one date, and a reminder that the burden sits with whoever holds the weakest evidence.

↑ Back to the key points

— By Dr. Ingrid Voss