The most authoritative test in the world for recycled plastic does not look for one of the largest things that comes out of recycled plastic. That is not a scandal, and no one is hiding it. It is written into the test's own published design.
To clear recycled PET for food contact in the European Union, a recycling process has to pass what is called a challenge test. The process is fed plastic that has been deliberately spiked with a panel of stand-in chemicals — toluene, chlorobenzene, chloroform, benzophenone, methyl salicylate and a few others — and then measured to see how well the process strips them back out. The stand-ins are chosen, in the words of the European Food Safety Authority's own 2024 guidance, as "substances with different molecular mass and polarity representative of all possible contaminants of concern."1 They model the things a previous owner might have done to a bottle: stored white spirit in it, kept solvent in it, misused it. The test asks, in effect, how good is this process at scrubbing out what the last person put in? It answers that question well. It is one of the most rigorous material-safety regimes anywhere.
The word "oligomer" does not appear in that selection logic. And the oligomers are a major class of what comes off recycled PET — in some studies the dominant non-targeted fraction migrating into food.
That gap — between what the gate checks and what actually comes out the other side — is the whole story. It is not the story that "recycled plastic is poison." It is something quieter and harder to dismiss: that the green word on the front of the pack, the one a careful shopper reaches for on purpose, is an origin story being read as a safety clearance, and that outside the one regulated food-contact channel — at ordinary shelf level — there is no reliable way for that shopper to find out whether any chemical gate touched the material at all.
The number that was wrong twice
The cleanest way into this is the story most people already half-remember, because it teaches the one reading skill the rest of it depends on.
In late 2024, a study in the journal Chemosphere set off a wave of "throw out your black plastic spatula" coverage. Researchers at the advocacy group Toxic-Free Future, led by Megan Liu, had analysed 203 black-plastic household products and found brominated flame retardants — chemicals engineered into electronics casings — turning up in kitchenware that touches hot food. One black sushi tray contained 11,900 parts per million of decaBDE; the level of one flame retardant, BDE-209, ran up to roughly 1,200 times the European limit of 10 ppm for that substance.2 The headline figure, though, was an estimated daily intake from a contaminated utensil: 34,700 nanograms a day.
That number was wrong. It was corrected twice.
The first correction, in December 2024, fixed a missing zero. The authors had compared their intake estimate against the US Environmental Protection Agency's reference dose for a 60-kilogram adult, and they had written that dose as 42,000 ng/day instead of 420,000 — a factor of ten.3 The second correction came on 3 July 2025, and it went further. The authors had misread the formula they used to estimate exposure in the first place. The corrected intake was not 34,700 ng/day. It was about 7,900.3
What that means deserves precision, because the temptation is to pick a side. The corrected figure, roughly 7,900 ng/day, sits comfortably below the adult reference dose. An industrial chemist, Mark Jones, wrote a published letter arguing that even the second correction did not go far enough and that the paper's "significantly contaminate" language should be softened.3 A reasonable reader could conclude the alarm was overstated and the system self-corrected, which is exactly what it did.
But here is the thing that did not change in either correction: the flame retardant was still in the kitchenware. The arithmetic that moved was an exposure example in the discussion section. The finding — that a chemical designed for a 1990s television was now in a slotted spoon — was never retracted. According to Retraction Watch, Liu said the correction touched "one exposure example in the discussion section of the study."3
So the reading skill is this: when a recycled-content chemical claim reaches you, read the measured exposure, not the headline, and check whether the number you are about to act on has since been corrected. Do that, and the spatula story stops being about a scary dose. It becomes about how the flame retardant got into the spoon in the first place — and why nothing on the spoon could have told you.
How a television became a serving spoon
DecaBDE was not an accident in that plastic. It was a feature, somewhere else. For years it was the dominant flame retardant in the rigid casings of electronics — the high-impact polystyrene of a television or a monitor — where its job was to slow a fire. The European Union banned it from new electronics in 2008; the US EPA phased it out by 2013; in 2017 it was added to Annex A of the Stockholm Convention, the global treaty that targets persistent organic pollutants for elimination.4 A chemical we had collectively decided to stop making.
The casing it lived in, though, outlived the decision. When that television was scrapped, its black plastic entered the recycling stream — and ran straight into a physical limit that almost no consumer knows about. Most plants sort plastic by shining near-infrared light on it and reading the polymer type from the reflected signal. Carbon black, the pigment that makes the plastic black, absorbs that light across the band and turns it into heat. The reflected signal is effectively zero. To the sorting machine, a black plastic object is invisible — which is why vendors now sell mid-wave-infrared scanners specifically to see the plastic the standard line cannot.5 A black casing cannot be identified, so it cannot be routed to its correct, clean stream. It goes into a blind mixed-plastics flow where an old electronics housing and a future kitchen utensil are not separated.
That is the pathway, and it helps to name what kind of claim it is. The flame retardant being present in the kitchenware is a measured fact about the material. Whether it migrates into your food is a separate question, governed by heat and time and fat. Whether that migration reaches a dose that matters is a third question again. The spatula study measured the chemical and modelled an exposure for a 60-kilogram adult. That adult is the least-exposed case there is.
Who the corrected number lands on
The reference dose the whole spatula correction turned on is calculated for a grown adult. But brominated flame retardants are persistent, fat-loving compounds that build up in the body over years and concentrate in lipid — including breast milk and house dust, the two routes that matter most for a small child.
Australian researchers who measured serum levels across ages found the body burden peaks early. The concentration of the four most common PBDEs rose from infancy and peaked in children aged about two and a half to three years. At that peak it sat at roughly 51 nanograms per gram of lipid. By adulthood — ages 31 to 45 — it had fallen to about 9.9.6 That is the toddler carrying around four to five times the adult body burden, with some studies putting the gap as high as tenfold. The US Agency for Toxic Substances and Disease Registry attributes this to the obvious mechanics of being small: breastfed infants take in the highest dose by body weight, and toddlers put dusty hands in their mouths.6
None of that makes 7,900 ng/day a dangerous dose. The point is narrower and harder to wave away: the calculation that reassured everyone modelled the person least likely to be exposed, and the people most likely to be exposed are the ones a parent is buying for. That is not an argument that the spatula is poisoning children. It is an argument that the reassurance was answering a different question than the one a parent is actually asking.
The reactor, not the conveyor
Here is where the spatula stops being the story and starts being an example of it.
The intuitive picture of recycling is a conveyor belt: old material goes in, gets cleaned, comes out as new material, and any unwanted chemicals are passengers that either get scrubbed off or ride along. On that picture, the only question is which banned chemicals hitched a ride and how much. That is the picture the spatula war was fought on — a list of named contaminants, measured against a limit.
Recycling is not a conveyor. It is a reactor. The heat and mechanical shear of melting and re-extruding plastic physically break the polymer chains. Researchers at Aarhus University tested seven PET food trays made with rising amounts of recycled content — zero, then 73, 90 and 100 percent. They found that the trays' own polymer was shedding short fragments of itself, cyclic oligomers, into the food simulant. As the recycled content rose, more of these fragments migrated — though the effect tracked with both the recycled-to-virgin ratio and the way each tray was extruded, not a clean one-to-one law. Analysis of the molecular weight showed the recycled plastic had been broken down further than the virgin — the physical signature of a chain cut shorter each time round the loop.7
These fragments are not exotic contaminants from somewhere else. They are pieces of the plastic itself, generated by the recycling. And they are one of the largest migrating fractions that comes off recycled PET — in some studies the dominant non-targeted species reaching the food. Yet there is no regulatory migration limit for PET oligomers, and — this is the candid part — the toxicological picture is thin. A Swiss-led 2023 evidence map reviewed the hazard and exposure evidence for 34 PET oligomers and found substantial gaps preventing proper risk assessment; as its summary puts it, "toxicological data is generally lacking" for these compounds, "making a risk assessment difficult."8
This is the catch, and it is essential to be exact about what it is and is not.
It is not a claim that the oligomers are harmful. The toxicological data are limited and incomplete, so the risk is genuinely uncertain — not shown safe, not shown harmful. To treat "unassessed" as "unsafe" would be to commit the same overreach the spatula headline did.
What it is: one of the largest things that migrates out of recycled PET is a substance with no name on any banned list, no migration limit, and no test pointed at it — not even in the lane that works. Go back to the EFSA challenge test. It screens for the surrogate misuse contaminants it was built to screen for, and it is excellent at that. It was never built to look at the fragments the recycling itself creates, because those are not what it is for. The gold standard is doing exactly what it was designed to do. What it was designed to do does not include a major part of what actually migrates.
That is the difference between "we checked and it's fine" and "we checked the thing we know how to check." The shopper reaching for the recycled tray hears the first. The test only supports the second.
The lane that works
It would be easy, and wrong, to let that slide into "recycled is dangerous." So it has to be said plainly, because it is true: there is a lane where the system genuinely works, and it works because someone deliberately built a gate into it.
That lane is closed-loop, food-grade recycled PET. Where the input is controlled, recycled PET is near-virgin in chemical safety for the contaminants the system targets. Controlled means a bottle-to-bottle deposit-return scheme that keeps a single, clean polymer stream, feeding a process authorised by EFSA under Regulation (EU) 2022/1616.1,9 Two long-running schemes show what that produces. Germany's deposit-return scheme has run since 2003; Norway's modern single-use deposit system has operated since 1999. Both yield high-purity, single-polymer, food-grade-segregated feedstock with high return rates. It is the closest thing to a chemical passport that exists for recycled plastic, and it exists because the input was controlled and the gate was built on purpose.9
So the concern here is not recycling. It is the absence of a gate everywhere the gate was not deliberately built — and the fact that the same word, "recycled," covers the lane with the gate and every lane without it, with nothing on the product to tell them apart.
This has happened before, and the fix is known
The shape of this failure is not new. It is at least forty years old, and the last time we watched it run its course, the hazard was not a chemical at all.
In 1992, residents in a number of Taiwanese apartment buildings discovered something strange: elevated gamma radiation coming from the walls of their own homes. The investigation that followed found the source in the steel. In late 1982 and 1983, a steel mill in northern Taiwan had melted scrap that included orphaned Cobalt-60 radioactive sources, and the contaminated metal had been rolled into reinforcing bar and built into homes and schools. By the time it was found, peer-reviewed dose reconstructions identified more than 200 buildings and roughly a thousand apartments, all constructed in 1982 to 1984, all built from steel from that mill.10 The radioactivity had been living in the concrete, invisibly, for a decade.
The same loop reached a shop — through a second Taiwanese mill, in a separate incident. Contaminated steel from Taiyang Steel Company, melted in early 1983, was fabricated into pipe fittings, shipped to the United States, and distributed by Ace Hardware. On 29 August 1984, health physicists at GA Technologies in California detected the radioactivity — after a plumber had already installed one of the fittings.11 A hazard rode a recycling loop, invisibly, into a hardware-store shelf and a family's plumbing, and was caught only by accident.
This is the same archetype as the spatula, forty years earlier: a hazard riding a closed recycling loop into an innocent everyday object, undetected, because nothing in the loop was watching for it. The only difference is that the hazard was a gamma-emitter instead of a brominated molecule. And the scrap-metal industry's response is the most useful part of the story. After decades of these incidents — the International Atomic Energy Agency reports on the order of 150 serious scrap-metal radiation events a year — the industry installed radiation portal monitors at scrap intake.12 They are imperfect; a well-shielded source can slip past one. But they exist because someone decided the loop needed a gate.
That is the entire hopeful thrust of this. The lanes that are safe are the lanes where a gate was deliberately built. The radiation portal monitor and the EFSA challenge test are the same idea: a deliberate check placed inside a loop that would otherwise carry whatever it was fed. The problem has never been that we cannot build the gate. It is that we build it only where we decide to, and then label everything else with the same green word.
The body against skin
Most of this has been about plastic touching food. But the cascade does not stop at the kitchen, and the next step lands directly on a child.
When a recycled PET bottle is not clean enough or not segregated for the food lane, it does not vanish. It is downcycled — diverted into applications where no chemical gate applies at all. One of those applications is clothing. Recycled polyester, the recycled-content fabric in so many sportswear and base-layer products, is frequently made from exactly this bottle feedstock.
In 2024, Czech researchers measured bisphenols in 54 everyday garments — 33 made from recycled material, 24 conventional. The recycled textiles carried a median BPA level of 13.5 nanograms per gram. The conventional fabric carried 7.66 — almost half as much.13 On its own that is a roughly two-fold difference, and for dry clothing the researchers' modelled skin exposure stayed below EFSA's tolerable daily intake for BPA.
The number that does not stay below it is the wet one. When the researchers simulated sweat — the warm, damp contact of a body in a base layer — exposure from the wet textiles often exceeded the tolerable intake. For the worst single sample, a recycled T-shirt containing both recycled cotton and recycled polyester, the modelled exposure under sweat simulation exceeded EFSA's BPA limit by somewhere between 125 and 570 times, depending on the absorption assumption.13
Every qualifier in that sentence matters and none of them should be stripped off. It is one sample, the worst one. It is sweat simulation, not a body. The absorption is modelled, not measured. The typical recycled garment in that study was about twice the conventional, not 570 times. But the point survives the qualifiers, because the point is not the multiplier. The point is the condition. The migration data that exists for these materials — Aarhus's oligomer trays at 40°C, the textile sweat simulation, the bottle studies — is the data the test conditions allowed. A child sweating into a base layer through a Saturday-morning football match is a system no controlled test on file actually modelled. The conditions Sarah will subject the product to are precisely the conditions the data does not cover.
This investigation continues below.
Want the next one in your inbox?
Weight, not what's in it
So why does the gap persist, when the world's regulators are themselves not yet sure how to verify these materials? Because of what the word "recycled" is actually counting.
Every instrument that creates demand for recycled plastic measures it by weight. Take the EU's Packaging and Packaging Waste Regulation, in force since February 2025. It sets recycled-content minimums for 2030: 30 percent for PET drinks bottles, 35 percent for other plastic packaging. Both are expressed as a percentage of plastic by weight.14 The UK's Plastic Packaging Tax works the same way. It charges packaging that contains less than 30 percent recycled plastic, again by weight; hit 30 percent by mass and the tax disappears, with no chemical condition attached.15 The policy intent is entirely sound — it is meant to pull virgin plastic out of the system, and it does. But a target denominated in weight has no term in it for chemistry. "30 percent recycled" is a mass statement. It tells you nothing about which lane the material came from or whether any gate touched it. The gap is in the denominator.
And the claim is about to drift even further from the molecule. From 2027, the UK's Plastic Packaging Tax is set to allow chemically recycled plastic to count toward the 30 percent threshold through a mass-balance approach — an accounting method that allocates recycled content across a batch of output products on the books, rather than tracing it to any one item. Under mass balance, as the recognised standards that define it spell out, recycled or renewable content is assigned to products independent of which physical molecules actually ended up in each one.15,18 In plain terms: a product can carry a "30 percent recycled" claim that is a bookkeeping credit, true on paper, while the actual molecules in that particular item are something else. The claim need not refer to the material in your hand.
This is the heart of why the shelf cannot answer the question. "Recycled," "rPET," "made from ocean plastic" — these are origin stories, and increasingly accounting entries. They are read as composition. They were never composition.
There is a real cost difference underneath this, and it is the quiet enforcement mechanism. Food-grade recycled PET — the gated lane, with its deposit-return feedstock and its EFSA-authorised process — costs materially more than virgin, because the gate is what costs. Outside the food lane, where no gate is required, the rational buyer sources the cheapest, least-tested recyclate available. The money saved by skipping the chemical check is, in a body, the antimony and the flame retardant and the fragment that stayed in.
The thing that's in all of it
One more substance complicates the tidy version of this story, and it cuts against the report's own simplest narrative, so it has to be put on the table directly.
Antimony — specifically antimony trioxide — is not a contaminant that sneaks into recycled PET. It is the catalyst used to make virgin PET in the first place. The literature describes it as "the most important catalyst used in the process," and it becomes incorporated into the polymer itself. Both the US EPA and the EU classify it as a priority pollutant. Total antimony of 100 to 400 milligrams per kilogram has been reported in PET containers regardless of brand.16 The food-grade purification step that "re-cleans" recycled PET — solid-state polycondensation — is itself a catalyst-driven reaction. It restores the polymer's molecular weight, but it does not appear to remove the antimony, because the antimony is part of how the reaction works.16
The literature is genuinely split on whether recycled PET carries more antimony than virgin — some studies report higher, others comparable — so the honest framing is not "recycled is worse." It is that antimony is in all of it, virgin and recycled alike, including the gated food-grade lane, by design. Antimony migration rises with heat and time; some hot-storage bottle studies report exceedances of the lower European drinking-water limit under severe conditions, while the 0.04 milligrams per litre food-contact migration limit is a separate, higher threshold.16 The point for a shopper is the one that keeps recurring: this is not a problem you solve by fleeing to virgin plastic. Virgin carries the catalyst too. The answer is not "buy new instead." The answer is the gated lane, and the ability to tell whether you are in it.
What would change this
Two things would materially change this assessment, and both deserve honest watching.
The first is disclosure. In 2026, intergovernmental bodies began circling a harmonised chemical-safety standard for recycled food-contact plastic — and that work is, today, only discussion, not adoption. If it reached adoption, the gap could close. The timing alone tells the story. Within roughly a month in the spring of 2026, the OECD published a report calling for the standard in April 2026; the FAO followed with its own on 13 May. The Codex Committee on Contaminants in Foods has the question on its agenda but has adopted nothing.17 Standard-setters asking, in the same season, for the same rule is the cleanest possible proof that the rule does not yet exist. If that work produced a label term that let a shopper tell a gated lane from an un-gated one at the shelf, this report's central claim — that she cannot find out — would no longer hold. The OECD's own framing is instructive: it notes that recycled plastics may contain "over 13,000 intentionally and non-intentionally added substances" and that "a lack of international harmonized standards regarding the chemical safety of recycled plastic results in inconsistent enforcement."17 That is a regulator describing the gap, not an advocate inventing it.
The second is the chemistry itself. The case here rests on the oligomers being both a major migrating fraction and outside the gate's scope. If replication showed that the food-grade challenge test does, in fact, capture the process-generated fragments — that the gate sees what it was thought not to see — the engine of this report would weaken to the point of needing rebuilding. As of the 2026 evidence, neither the disclosure nor the chemistry has moved. Both are the right things to watch.
The strongest version of the opposing case deserves to be stated in full, because it is reasonable: the food-grade lane is among the most heavily regulated material streams in the world; the spatula scare was overstated by an order of magnitude and corrected in public by the authors themselves; the oligomers are unassessed, not shown to be harmful; the dramatic figures come from conservative test conditions and worst-case samples; and recycling is a genuine environmental good that an anxious consumer should not abandon for virgin plastic. Every clause of that is true, and this report concedes all of it. The one thing it does not account for is the thing this whole piece is about: none of it tells a shopper which lane she bought, and the world's own regulators are, right now, asking for the rule that would.
What to do on Saturday
So, standing in the shop, with a recycled-content product in your hand and no way to read its biography off the pack — what is actually worth doing?
For anything that holds food or drink, the meaningful distinction is the gated lane. Look for explicit food-contact recycled PET that names its authorisation — wording that references EFSA Regulation (EU) 2022/1616 for European products, or an FDA process review for American ones, ideally with bottle-to-bottle or deposit-return feedstock named. That is the lane with the gate. A bare "made from recycled plastic" or "contains 30% recycled content" on a food product is a weight claim, not a safety one, and after 2027 in the UK it may not even describe the molecules present. If the brand's page does not name the food-contact authorisation, the question to send their customer-service line is specific: Is this food-contact recycled PET authorised under Regulation (EU) 2022/1616, and is the feedstock closed-loop? The brands in the gated lane can answer it in a sentence. The ones that go quiet have told you which lane you are in.
For recycled-polyester clothing — base layers, sportswear, anything worn against sweat — there is no equivalent gate to look for, because none exists. The honest reading is that "recycled polyester" is an environmental claim about the bottle it came from, not a chemical claim about what it does on skin, and the wet-contact data is the data we have least of. That is not a reason to panic; the typical recycled garment in the one study we have ran about twice the conventional, not hundreds of times. It is a reason not to read the recycled label as a reassurance it was never making.
And for the black plastic kitchen utensil that started this — the slotted spoon, the spatula — the reading is simplest of all. Black plastic is the one stream the sorting machines cannot see, which means it is the stream most likely to carry whatever was fed into it. A utensil in a polymer the line can actually identify is the safer bet: a wooden spoon, stainless steel, or a pale plastic the scanner can read and route. None of this requires fleeing to virgin plastic — virgin carries antimony too — and none of it is a reason to stop choosing recycled where the gate is real. It is a reason to know that the green word is an origin story, and to ask, of the products that matter most, which lane it came from. Right now, for most of what is on the shelf, the honest answer is that you cannot find out. A harmonised, ordinary shelf-level label that consistently discloses recycled-content lane, feedstock control, and chemical gate is not yet available.