In January 2009, the US Environmental Protection Agency issued a consent order for two new chemicals. The premanufacture notices -- PMN P-08-0508 and P-08-0509 -- had been filed by DuPont the previous year. The compounds were processing aids for fluoropolymer manufacture. The consent order approved their production under conditions: 99 per cent capture of discharges and emissions, respirators with a protection factor of 3,000 for exposed workers.
Buried in the consent order's language was a phrase that would come to define the next fifteen years of cookware chemistry: the information DuPont submitted was, in the EPA's own assessment, "insufficient to permit a reasoned evaluation of the health and environmental effects."1
The compounds were approved anyway.
Those two chemicals are now known as GenX. They replaced PFOA -- the compound that made Teflon possible and made headlines when its health effects could no longer be contained. Today, GenX residuals sit in the coating of pans labelled "PFOA-free" on shelves across every major market. The label answers a question. It does not answer the right one.
The question a consumer cannot currently answer: does the regulatory system that approved your pan's coating evaluate the property that makes PFAS permanent?
The Label
Start with what the consumer sees. A non-stick pan in a shop. Somewhere on the packaging, two words: "PFOA-free."
Try to determine what those words mean in regulatory terms. Search the US Code of Federal Regulations. Search FDA guidance documents. Search FTC enforcement advisories. Search EU REACH annexes, UK retained legislation, Australian and Canadian product safety standards.
The term "PFOA-free" has no legal definition in any jurisdiction.2
No federal agency in the United States defines what a manufacturer must demonstrate to print those words. The EU restricts PFOA itself under the Persistent Organic Pollutants Regulation 2019/1021, but "PFOA-free" as a marketing claim is not defined, standardised, or policed.3 A manufacturer can print "PFOA-free" on cookware that contains dozens of other PFAS compounds, including GenX, PFBS, and the fluoropolymer PTFE itself -- and violate no regulation in doing so.
One state has addressed this. California AB 1200, effective January 2024 for cookware, prohibits manufacturers from claiming a product is "free of" a specific chemical if any chemical from the same hazardous family is intentionally present.4 As of March 2026 -- over two years since the cookware provisions took effect -- no public enforcement action against a cookware manufacturer has been identified. The California Attorney General announced in October 2023 that enforcement would be a "priority."4 The priority has not produced a visible result.
This is the first clue. The words on the label are not backed by the machinery of law. "PFOA-free" is a compound-specific absence claim operating in a class-persistence world, and no one in the regulatory architecture is required to verify whether the claim tells the consumer anything meaningful about what is in the pan.
The Gate
The assumption most consumers carry: if a replacement chemical is in a product on the shelf, someone assessed whether it was safe before it got there. The FDA's food contact framework is the presumed gate. A manufacturer submits evidence. The agency reviews it. The substance is approved for food contact use.
This is how PTFE -- the polymer that is the non-stick coating -- entered the regulatory system. Under 21 CFR 177.1550, first codified on 29 September 1978, perfluorocarbon resins are authorised for food contact use as coatings, subject to extraction limits: total extractables may not exceed 3.1 mg/dm2, and fluoride extractables may not exceed 0.46 mg/dm2.5 The regulation was designed for the finished polymer. It tests what leaches from the coating under controlled conditions: two hours at reflux with water, ethanol, heptane, and ethyl acetate.
GenX did not come through this gate.
A search of FDA's Food Contact Substance Notification System for HFPO-DA (the chemical name for GenX, CAS 62037-80-3) returns no matching entries. A broader notification covering GenX as part of a larger category cannot be definitively ruled out.6 GenX is not in the food contact database. It entered commerce through an entirely different door: EPA's Toxic Substances Control Act, Section 5(e) -- the new chemical program for manufacturing substances.1
The distinction is architectural, and it matters. The FDA framework governs substances intended to contact food. GenX is classified as a processing aid -- used in the manufacture of PTFE, not intended to be in the finished product. But processing aid residuals remain in the finished coating. They contact food during cooking. Academic migration studies have detected PFAS leaching from cookware at temperatures as low as 70C with acidic food simulants -- well below any temperature warning threshold.7
The compound contacts food. Neither agency assessed it for that purpose. The EPA consent order regulated manufacturing emissions. The FDA regulation tests the finished polymer. GenX residuals in the coating sit in the gap between the two frameworks. One agency approved it as a manufacturing chemical. The other agency regulates the product it ends up in. The scope boundary between EPA's TSCA authority and FDA's food contact authority is where the consumer's question disappears.
It takes a certain precision to build a regulatory architecture where a compound can be approved for manufacturing, leave residuals in a product that touches food, and fall outside the assessment scope of both the manufacturing authority and the food safety authority. This is not negligence. The frameworks were designed for different purposes in different decades. 21 CFR 177.1550 was written in 1978, when the processing aids used in fluoropolymer manufacture were not the subject of toxicological concern. The TSCA new-chemical program assesses manufacturing risk, not dietary exposure. Neither was designed to catch GenX. Neither did.
The Numbers
Twelve years after the consent order approved GenX with "insufficient" evidence, the EPA's Office of Research and Development published a toxicity assessment.8
The findings, released 25 October 2021: GenX's chronic oral reference dose is 3 nanograms per kilogram of body weight per day. For comparison, the EPA established PFOA's reference dose at 20 nanograms per kilogram per day in May 2016.9
The replacement's safety threshold is 6.7 times more restrictive than the compound it replaced.
A note on what this means and what it does not. A reference dose is not a direct toxicity ranking. The GenX RfD of 3 ng/kg/day incorporates a composite uncertainty factor of 3,000 -- ten for intraspecies variation, three for interspecies extrapolation, ten for subchronic-to-chronic duration, and ten for database deficiency.8 That last factor -- database deficiency -- is the EPA's formal acknowledgment that the evidence base for GenX is incomplete. The 2018 draft RfD for GenX was 80 ng/kg/day. The final value dropped by a factor of 26.7 from the draft, primarily because the EPA applied additional uncertainty factors as data gaps became apparent, not because a single new study proved greater toxicity.8
So the inversion is real, but not in the way a headline might suggest. The replacement does not have a stricter safety threshold because toxicologists proved it is more dangerous. It has a stricter safety threshold because the regulatory system knows so little about it that the agency must assume the worst. The uncertainty is itself the product of the architecture. The compound was approved in 2009 with insufficient data. The data deficiency was quantified in 2021 as a 3,000-fold uncertainty factor. The gap between approval and assessment is twelve years. Twelve years of commercial use, twelve years of processing-aid residuals in cookware coatings, twelve years of a label that says "PFOA-free" while the replacement compound's safety profile remained unresolved.
The FDA food contact framework for PTFE (21 CFR 177.1550) was never updated to account for the EPA's GenX findings. The two assessments -- FDA's 1978 polymer authorisation and EPA's 2021 toxicity assessment -- sit in separate databases with no mechanism to reconcile them. One agency approved the product. Another agency found the processing aid needs more caution than the compound it replaced. Both are correct within their respective mandates. The architecture does not require them to talk to each other.
The Clock
PFOA is not the first compound to take this ride.
Tetraethyl lead was added to petrol in 1923. Refinery workers died in 1924. The evidence of neurotoxicity accumulated for decades. The US began its phase-out in 1973, completed it in 1996, and the global ban on leaded petrol was not achieved until 2021 -- ninety-seven years from the first worker deaths.10
DDT was introduced as a pesticide in the 1940s. When the EPA restricted it in 1972, the agency's own "Project Safeguard" trained farmers to use highly toxic organophosphate substitutes.11 The restricted compound was replaced. The replacement was not safer. The system moved on.
BPA in plastics followed the same arc. "BPA-free" products appeared across the market. Many contained BPS or BPF -- structural analogues with comparable estrogenic activity. In December 2024, the EU adopted Regulation 2024/3190, banning not just BPA but an entire class of hazardous bisphenols in food contact materials.12 It took over two decades of compound-specific BPA restriction failing before the regulatory system moved to a class-based approach.
Asbestos is the most instructive. Chrysotile -- responsible for 95 per cent of commercial asbestos use -- was exempted from restrictions for decades while amphibole fibre types were banned. The regulatory system applied compound-by-compound thinking to fibre types. The United States did not ban chrysotile until March 2024.13
The academic literature has a name for this output: regrettable substitution. The term entered the scientific vocabulary through Zimmerman and Anastas's 2015 paper in Science, which argued for designing safer alternatives from the outset rather than replacing hazardous compounds with structural analogues whose safety profiles are unknown.14 Kwiatkowski et al., writing in Environmental Science & Technology Letters in 2020 with fifteen co-authors including a former director of the National Institute of Environmental Health Sciences, made the systematic case for managing PFAS as a chemical class. Their argument: environmental persistence is conferred by the carbon-fluorine bond, which is present in all 12,000-plus PFAS compounds. Managing them one at a time, they wrote, "has not been effective at controlling widespread exposure to this large group of chemicals with known and potential hazards."15
The word "substitute" in the regulatory vocabulary means something different from what consumers hear. Consumers hear "safer alternative." The regulatory architecture produces something else: the next rotation. The restricted compound exits. The replacement compound enters the same approval gate with the same structural property -- the C-F bond -- and the same initial evidence deficit. The clock resets.
The Patent
The rotation has an economic engine.
DuPont filed its premanufacture notices for GenX in 2008 -- before PFOA was formally restricted in most jurisdictions. The EPA Stewardship Program's final deadline was 2015. The replacement was in the regulatory pipeline seven years before the original's phase-out was complete.1
On 1 July 2015 -- the same year the Stewardship Program deadline was met -- DuPont completed the spinoff of Chemours as an independent publicly traded company. The distribution: one share of Chemours common stock for every five shares of DuPont. Chemours inherited GenX intellectual property, the fluoroproducts business, and approximately two-thirds of DuPont's environmental liabilities, despite representing only 19 per cent of DuPont's revenue.16
Cordner et al., writing in Environmental Science & Policy in 2025, examined GenX specifically and concluded that the substitution was "a foreseeable product of structural factors, including corporate influence on science and regulation, that lead to favorable outcomes for industry to the detriment of the broader public."17 The finding is structural: the entity with the deepest expertise in fluorine chemistry -- the entity that manufactured the restricted compound for decades -- is the entity best positioned to develop, patent, and sell the replacement. Under compound-by-compound regulation, each restriction event is simultaneously a market-creation event for the incumbent.
The pattern is architectural. It is not specific to DuPont or Chemours. It applies to any entity operating within a compound-by-compound regulatory framework where the restricted substance belongs to a chemical class and the replacement comes from the same class. The restriction creates a replacement market. The incumbent enters first. The regulatory clock resets. The label changes. The C-F bond does not.
In 2023, DuPont, Chemours, and Corteva agreed to a $1.185 billion settlement with US water systems, with a broader $4 billion cost-sharing arrangement over up to twenty years.18 Chemours's Advanced Performance Materials segment -- which includes fluoropolymers -- generated $1.3 billion in net sales in 2025 alone.19 The settlement costs are substantial but manageable against the ongoing revenue from the very chemistry the settlement addresses.
The Defence
In May 2024, Meyer Corporation and Groupe SEB -- manufacturer of Tefal -- launched the Cookware Sustainability Alliance, a 501(c)(6) non-profit. Founding members included SCANPAN and Tramontina, with additional brands All-Clad and Farberware under the alliance.20
According to Chemical & Engineering News reporting, CSA spent at least $218,000 lobbying in California, at least $15,000 in Vermont, and at least $25,000 in Connecticut.21 The alliance was active in thirteen states where PFAS legislation was being considered. In California, Governor Newsom vetoed SB 682 -- a bill that would have banned PFAS in cookware -- in October 2025.21 In Vermont, the alliance convinced legislators to delay implementing the state's cookware ban by several years, to 2028.21
In France, the campaign took a different form. Groupe SEB organised employees to rally outside the National Assembly under the slogan "Touche pas a ma poele" -- "Don't touch my pan." The company hired Rivington, a public affairs consulting firm, to secure cookware exemptions from the national PFAS ban.22 Tefal ran billboard campaigns claiming "Our non-stick coatings are recognized as safe."23 The Federation of European Cookware, Cutlery and Houseware Industries placed a paid editorial in Politico asserting that fluoropolymers are "already known to be safe across their whole lifecycle."23
France adopted its PFAS ban -- Law No. 2025-188 -- on 20 February 2025, by a vote of 231 to 51. Cookware was exempted from the final text.24 The ban applies to cosmetics, wax products, consumer textile clothing, and footwear from January 2026. The item the lobbying campaign was built to protect -- the non-stick pan -- was removed.
None of this is unlawful. Lobbying is constitutionally protected in the United States and a recognised part of the legislative process in France and the EU. The documented expenditures and campaigns represent lawful political participation by an industry defending its commercial interests through democratic channels.
The structural observation is different from a moral one. The compound-by-compound architecture creates a replacement market. The replacement market generates revenue. Revenue funds the defence of the architecture that generates it. Each restriction event creates a patented replacement for the incumbent. Each class-wide restriction proposal threatens the entire cycle. The lobbying target is not any specific compound -- PFOA is already gone. The target is the architectural redesign that would end the replacement cycle itself.
The Precedent
Class-wide restriction has been done before.
PCBs -- polychlorinated biphenyls -- were manufactured commercially from 1929 to 1977 in the United States, almost exclusively by Monsanto under the commercial name Aroclors.25 Congress singled out PCBs for a class-wide ban under TSCA in 1976, effective 1979. The Stockholm Convention on Persistent Organic Pollutants listed PCBs among its initial "dirty dozen" in 2001.26 All 209 PCB congeners were restricted -- not individually assessed, not evaluated compound by compound, but banned as a class on the basis of shared persistence and bioaccumulation.
The conditions that made the PCB ban achievable are specific and instructive. Monsanto was the sole US manufacturer -- a single company's exit could eliminate the class. PCBs had limited commercial diversity: 209 congeners in identifiable industrial applications, with no consumer-facing product identity. There was no "PCB-free" marketing to provide consumers with cognitive closure. And critically, no adjacent consumer product depended on PCBs for a function the public had come to treat as essential.
PFAS lack every one of these conditions. Dozens of manufacturers produce PFAS compounds globally. The class encompasses over 12,000 compounds across sectors from semiconductors to medical devices to cookware. The non-stick pan has consumer identity -- domestic, familiar, defended. And fewer than one per cent of all PFAS have been individually tested, according to Kwiatkowski et al.15 -- meaning compound-by-compound assessment will structurally never cover the class.
The EU's bisphenol class ban (Regulation 2024/3190, effective January 2025) demonstrates that class-based restriction of food contact chemicals is achievable.12 It took over two decades of BPA-specific restriction failing -- and the emergence of BPS and BPF with equivalent endocrine effects -- before the system moved to a class approach. The regulatory system can learn. The learning takes decades.
The ECHA universal PFAS restriction proposal, submitted in January 2023 by Denmark, Germany, the Netherlands, Norway, and Sweden, is the most ambitious class-wide chemical regulation ever attempted.27 The August 2025 update explicitly found that substitutes are available for consumer nonstick cookware -- meaning cookware does not qualify for the 13.5-year derogations proposed for semiconductors and energy technology.27 RAC and SEAC final opinions are expected by the end of 2026. Implementation would follow with transition periods. The proposal was submitted over three years ago. It may reach implementation by 2028 at the earliest. During that interval, the architecture it proposes to replace continues to operate.
The system can self-correct. The question the PCB precedent poses is not whether class-wide restriction is possible. It is how many decades the carousel turns before the correction arrives -- and whether the lobbying infrastructure that has already secured exemptions in France, delayed implementation in Vermont, and defeated legislation in California can extend the timeline further.
The Architecture
Pull back far enough and the mechanism comes into view.
The academic literature calls the output regrettable substitution.14 What it does not yet name is the mechanism that makes regrettable substitution the architecture's default output rather than an occasional failure. The structure has five interlocking components, each documented in the PFAS regulatory record.
The Approval Gate. A new compound enters the regulatory system with a minimal evidence base. It is assessed against data available at the time of submission -- which, by structural necessity, is less than will exist after years of independent research. The EPA's consent order for GenX is the document: approved in January 2009 with evidence the agency itself described as "insufficient."1
The Evidence Accumulation. Independent researchers, government agencies, and biomonitoring programs build a toxicological profile over years to decades. For GenX, this produced the EPA's October 2021 toxicity assessment -- twelve years after the consent order -- establishing a reference dose 6.7 times more restrictive than PFOA's.8
The Restriction Trigger. When public evidence reaches a threshold sufficient to trigger regulatory action, the compound is restricted. The restriction is experienced by consumers as the system working. PFOA's phase-out under the Stewardship Program (completed 2015) registered as progress. "PFOA-free" appeared on labels. The question was closed.
The Clock Reset. During the evidence accumulation for the restricted compound, the replacement has already been developed and submitted. DuPont filed the GenX PMN in 2008 -- before the 2015 Stewardship deadline.1 The replacement enters the Approval Gate with its own fresh evidence deficit. The regulatory clock restarts at zero. GenX is measured in the NHANES biomonitoring program but detected in fewer than 0.4 per cent of blood specimens -- making it functionally invisible to the population-level surveillance that would accelerate the next Restriction Trigger.28
The Architecture Defence. Industry actors lobby to maintain the compound-by-compound framework itself. The CSA's formation, its multi-state campaigns, Groupe SEB's France exemption -- these are not actions to save PFOA. PFOA is gone. They are actions to preserve the structure that generates the replacement market. The compound-by-compound architecture is economically self-reinforcing: each restriction creates a patented opportunity for the incumbent, and each class-wide proposal threatens the cycle that produces those opportunities.
The mechanism -- the Class Substitution Loop -- does not require bad actors. It requires only two conditions: a compound-by-compound regulatory design premise and a chemical class with shared persistence. The C-F bond does not discriminate by chain length. Environmental persistence is a class property, not a compound property.15 The regulatory architecture was designed for a world where chemicals were individual substances with individual hazard profiles. The discovery that persistence is a class property made the design premise obsolete. The obsolete design continues to operate. It is defended because it is profitable.
Everyone followed the rules. The FDA approved PTFE within its statutory mandate. The EPA assessed GenX when independent evidence accumulated. ECHA is proposing class-wide restriction through proper channels. Groupe SEB lobbied through lawful means. The rules are the architecture. The architecture produces the carousel.
What Would Change This Analysis
The strongest counter-evidence would come from three directions.
First: if the ECHA universal PFAS restriction proposal is adopted without cookware exemptions and implemented within its proposed timeline, the compound-by-compound architecture would be structurally replaced for PFAS in food contact. This would not retroactively invalidate the Loop's operation over the past decades, but it would demonstrate that the architecture can be redesigned when institutional will and scientific consensus align. The ECHA August 2025 update's finding that cookware substitutes are available is evidence in this direction.27
Second: if future research demonstrates that cookware-specific PFAS exposure is negligible relative to drinking water and food packaging pathways, the consumer risk from this specific product category would be lower than the regulatory critique implies. No study has isolated cookware's quantitative contribution to total PFAS body burden. This report diagnoses an architectural flaw in how cookware coatings are regulated. It does not claim that cookware is the dominant PFAS exposure source. If cookware exposure is negligible, the architecture is still the architecture -- but the urgency of cookware-specific reform would diminish.
Third: if independent research finds that short-chain PFAS environmental persistence is significantly lower than long-chain PFAS -- that the C-F bond in shorter compounds does break down under environmental conditions -- the class-level persistence argument would weaken. Kwiatkowski et al.'s case for class-based management rests on the C-F bond as a universal class-level hazard indicator. If persistence varies meaningfully by structure rather than simply by half-life, compound-by-compound assessment might be more appropriate than this analysis suggests. Current evidence does not support this: the C-F bond is the strongest single bond in organic chemistry, and it does not discriminate by chain length.15
A note on what would not change this analysis. Evidence that individual PFAS compounds have different toxicity profiles does not address the Loop's structural claim. The mechanism operates on persistence, not toxicity. Different toxicity does not mean different persistence. The C-F bond is indifferent to the compound it anchors.
The Levers
For consumers seeking to evaluate their cookware through the regulatory architecture this report has mapped:
What "PFOA-free" tells you: One specific compound from a class of over 12,000 was not intentionally added. The term has no legal definition in any federal jurisdiction and does not require disclosure of which PFAS compounds are present.
What to look for instead: Cookware marketed as PTFE-free or fluoropolymer-free -- which addresses the class rather than the compound. Ceramic, cast iron, carbon steel, and stainless steel coatings do not use PFAS processing aids. The ECHA proposal's finding that substitutes are available for consumer cookware reflects the current state of alternatives.27
What is already changing: Minnesota's Amara's Law banned PFAS in cookware effective January 2025 -- the first US state to address the class rather than the compound.29 California AB 1200 prohibits misleading "free of" claims when the same chemical family is present.4 The EU bisphenol class ban demonstrates that class-based food contact regulation is implementable.12
What the regulatory architecture does not yet require: Disclosure of which specific PFAS compounds are in your cookware. Migration testing of finished products under real-world cooking conditions. Reconciliation between EPA manufacturing approvals and FDA food contact standards. Class-level assessment of persistence before compound-level approval for commerce.
The Gap
The pan is still in the shop. It still says "PFOA-free."
The regulatory system that approved the coating operates on a design premise from a time before the class concept existed for this chemistry. It evaluates compounds individually. The hazard -- environmental persistence conferred by the carbon-fluorine bond -- is a class property shared by all 12,000-plus PFAS. The label claims what is absent. The regulation does not require disclosure of what is present. The approval gate lets compounds through with evidence the agency itself calls insufficient, and the assessment that catches up twelve years later sits in a different database with no mechanism to revise the original approval.
The carousel has been turning for a century across chemical classes. TEL, DDT, asbestos fibre types, BPA, PFOA. Each restricted compound was once an approved substitute. Each "free from" label closed the consumer's question before the deeper question could form. The mechanism has five components, and none of them require anyone to act in bad faith. They require only a system designed for individual compounds and a reality in which the hazard operates as a class.
Something does not add up. Now you can see where the numbers diverge.
The architecture is the gap. The gap is visible. That is not nothing.