Correction

“Clean” Is a Retail Category, Not a Safety Standard

The claim

“‘Clean’ and ‘non-toxic’ on a pack mean something a regulator checks.”

The short answer

In the US, no — FDA does not define “organic” for cosmetics and has no definition of “clean” or “non-toxic”; MoCRA created none. In the EU there is an enforceable rule, and “free from” claims routinely fail it.

Evidence status

Settled

The Cosmetic Record · Published 2026-08-06 · Last checked 2026-08-06 · 21 sources · Corrections: none

1. What FDA says about “organic”, in FDA's own words

FDA answers this on a page of its own, in four questions. The first is “Does FDA have a definition for the term ‘organic’?” and the answer begins with the word No.

No. FDA regulates cosmetics under the authority of the Federal Food, Drug, and Cosmetic Act (FD&C Act) and the Fair Packaging and Labeling Act (FPLA). The term “organic” is not defined in either of these laws or the regulations that FDA enforces under their authority.

The definition that does exist belongs to a different department. USDA's Agricultural Marketing Service runs the National Organic Program, whose regulations define “organic” for agricultural ingredients and set labelling tiers by percentage. A cosmetic carrying that seal has satisfied an agriculture rule about how its plant material was grown, not a cosmetic rule — there is no cosmetic rule for the word.

The fourth question is the one that matters most, and FDA also answers it with No: “An ingredient's source does not determine its safety.” Its example is plants, which “whether or not they are organically grown, contain substances that may be toxic or allergenic”; under section 601(a) of the FD&C Act every cosmetic product and ingredient is subject to the same safety requirement regardless of provenance.¹

2. “Clean” and “non-toxic”: no definition, and MoCRA did not create one

FDA's cosmetics site carries a small cluster of pages on labelling claims. The navigation block repeated on each of them lists five: “Alcohol Free”, “Cosmeceutical”, “Cruelty Free”/“Not Tested on Animals”, “Hypoallergenic” and “Organic”.¹² There is no page for “clean”, none for “natural” and none for “non-toxic” — an absence rather than a statement, and section 8 records the limits of the search behind it.

The “Hypoallergenic” page is worth reading precisely because that word looks the most technical of the set. FDA's position, unchanged since 18 October 2000, is flat: “There are no Federal standards or definitions that govern the use of the term ‘hypoallergenic.’ The term means whatever a particular company wants it to mean.” The page explains why: FDA issued a regulation defining the term in 1975, and the US Court of Appeals for the District of Columbia struck it down.²

The Modernization of Cosmetics Regulation Act of 2022 did not change this. MoCRA's definitions sit at 21 U.S.C. 364, and there are five: adverse event, cosmetic product, facility, responsible person and serious adverse event.⁴ Not one is a marketing adjective. What MoCRA added was facility registration, product listing, adverse-event reporting and a duty to hold safety substantiation records — and on that duty FDA's own summary is explicit: “Neither the law nor FDA regulations require specific tests to demonstrate the safety of individual products or ingredients.”³

Two American instruments do bite on the words themselves. 21 U.S.C. 362(a) deems a cosmetic misbranded “if its labeling is false or misleading in any particular”.⁵ And the Federal Trade Commission's Green Guides carry a section on “free-of” claims and a section on “non-toxic” claims — the latter four sentences of guidance and one worked example.

(a) It is deceptive to misrepresent, directly or by implication, that a product, package, or service is non-toxic. Non-toxic claims should be clearly and prominently qualified to the extent necessary to avoid deception. (b) A non-toxic claim likely conveys that a product, package, or service is non-toxic both for humans and for the environment generally. Therefore, marketers making non-toxic claims should have competent and reliable scientific evidence that the product, package, or service is non-toxic for humans and for the environment or should clearly and prominently qualify their claims to avoid deception.

Their status is stated in the guides themselves: they “do not confer any rights on any person and do not operate to bind the FTC or the public”, and in any enforcement action the Commission “must prove that the challenged act or practice is unfair or deceptive in violation of Section 5 of the FTC Act”.⁶ That is a long way from a definition a regulator checks before a pack ships.

Something does put criteria behind the word “clean”, though — a retailer, for its own shelves. Sephora's published chemicals policy describes its programme as one that “highlights brands that voluntarily comply with the program's standards to formulate products without certain ingredients”, a “formulated without” list “now in its third version having continually updated the … list since launch in 2018”.²¹ Voluntary, retailer-set, and revised whenever the retailer decides: that is a merchandising category, and it is an accurate description of the word rather than a criticism of it.

3. The EU does have a rule, and it is stricter than people assume

The European parent provision is one sentence in the Cosmetics Regulation.

In the labelling, making available on the market and advertising of cosmetic products, text, names, trade marks, pictures and figurative or other signs shall not be used to imply that these products have characteristics or functions which they do not have.

Article 20(2) then instructed the Commission to adopt common criteria, and it did: Commission Regulation (EU) No 655/2013, applicable since 11 July 2013. Its annex sets six — legal compliance, truthfulness, evidential support, honesty, fairness, informed decision-making — and the responsible person must keep the substantiation in the product information file. The first of them is the one that catches a claim of absence:

Claims which convey the idea that a product has a specific benefit when this benefit is mere compliance with minimum legal requirements shall not be allowed.

How those criteria apply to the phrase “free from” was written up in 2017 by the Sub-Working Group on Claims, whose Technical document on cosmetic claims gained an Annex III on “free from” claims and an Annex IV on “hypoallergenic”, both applicable from 1 July 2019. The document is candid about its own weight, in capitals on its first page: it “is not a European Commission document” and the views in it “are not legally binding”; it calls itself a “tool” for case-by-case application by the Member States.⁹ It is guidance for national competent authorities rather than law — and it is the clearest published statement of how the six criteria bite, so the next three sections use its worked examples.

4. Worked example one: “paraben-free” does not tell you which paraben

“Paraben” is not one substance and the European annexes do not treat it as one. Five parabens are prohibited outright, at Annex II entries 1374 to 1378: isopropylparaben, isobutylparaben, phenylparaben, benzylparaben and pentylparaben, added by Commission Regulation (EU) No 358/2014. The reason given in recital 7 is not a finding of harm but an absence of a file — the SCCS “noted that limited or no information was submitted by industry” for their safety evaluation, so “the human risk cannot be evaluated”.¹⁰

Two more are capped rather than banned. Annex V entry 12 permits 4-hydroxybenzoic acid and its methyl and ethyl esters at 0,4 % (as acid) for a single ester and 0,8 % for mixtures of esters. Entry 12a covers butylparaben and propylparaben at 0,14 % (as acid) for the sum of the individual concentrations, and forbids them in leave-on products designed for the nappy area of children under three years of age.⁷ That 0.14% is a conversion, not a round number: Commission Regulation (EU) No 1004/2014 recorded the SCCS opinion SCCS/1514/13 concluding that use was safe as long as the sum did not exceed 0,19 % as esters, and recital 10 converted 0,19% as esters into “its equivalent as acid, 0,14 %”.¹¹

So a pack reading “paraben-free” collapses three different regulatory states — banned for want of data, capped at 0.4% as acid, and capped at 0.14% as acid with one product type excluded outright — into a single absence. The EU claims guidance goes further, and treats the phrase as unfair.

Certain parabens are safe when used in accordance to Regulation (EC) No 1223/2009. Considering the fact that all cosmetic products must be safe, the claim ‘free from parabens’ should not be accepted, because it is denigrating the entire group of parabens.

5. Where the paraben story came from, and what that paper actually did

Behind the phrase sits one 2004 paper — the paper the European Commission's consumer-products committee was asked to assess the following year¹⁷ — and it is worth describing by what it did rather than by what it is said to have shown. Darbre and colleagues measured individual parabens in tissue from 20 human breast tumours, supplied in collaboration with the Edinburgh Breast Unit Research Group,¹⁶ by high-pressure liquid chromatography followed by tandem mass spectrometry. The mean total was 20.6 ± 4.2 ng per gram of tissue; methylparaben — the paraben with the lowest oestrogenic activity in the in vitro (laboratory glassware) and in vivo (live animal) assays, since potency rises with the length of the alkyl side chain¹⁷ — was 62% of what was recovered. The stated purpose was to establish that parabens could be found intact in breast tissue at all.¹⁵ It ran in the same issue of the Journal of Applied Toxicology as a four-page commentary by Harvey and Everett on, among other things, “what can be inferred from this type of data (and what can not, such as the cause of these tumours)”.¹⁶

Four things the study did not contain are the whole of the matter: no healthy control tissue, no route of entry, no source for the parabens, and no test of causation. That is not our reading but the EU committee's, which reviewed the paper the following year and recorded both the criticisms and the authors' replies.

Darbre et al. acknowledge that the study on the concentrations of parabens in breast tumours could not identify either the route of entry or the source of the parabens. Carcinogenicity was not considered in the study and the presence of parabens was not claimed to cause the breast tumours.

The committee's conclusion, in its own words, was “that, in the light of the present knowledge, there is no evidence of demonstrable risk for the development of breast cancer caused by the use of paraben-containing underarm cosmetics”.¹⁷ The later European restrictions on parabens are real, and section 4 above sets them out — but they were made for missing safety data and for infant exposure under a nappy, not for this. A marketing claim built on the 2004 paper is not reporting the regulatory history; it is reporting a headline about it.

6. Worked example two: preservatives are a trade-off the regulators keep adjusting

The second reason a “free from” list ages badly is that the underlying table moves. Take methylisothiazolinone, Annex V entry 57. Commission Regulation (EU) 2016/1198 removed it from leave-on products entirely, recording the SCCS conclusion that “the concentration of 100 ppm of methylisothiazolinone in cosmetic products is not safe for the consumer”; compliance was required from 12 February 2017.¹² Just under a year later Commission Regulation (EU) 2017/1224, citing opinion SCCS/1557/15, cut the surviving rinse-off ceiling from 0,01 % to 0,0015 %.¹³ The 3:1 mixture with methylchloroisothiazolinone sits at entry 39, rinse-off only, also at 0,0015 %, and the consolidated text notes that the two entries “are mutually exclusive”.⁷

Phenoxyethanol is Annex V entry 29, capped at 1,0 %, and it is in a large share of the products the word “clean” is attached to: the survey described in section 7 found it in 591 of 1,470 ingredient lists — 40.2 %, the second commonest potential allergen after fragrances and botanicals.¹⁸ The claims guidance treats a “free from” claim about it exactly as it treats parabens: phenoxyethanol and triclosan “are safe when used according to the Cosmetics Regulation. Hence the claim free from these substances should not be accepted because it is denigrating authorised substances.”⁹

Formaldehyde is the sharpest case, because the pack can be literally true and still fail. The substance itself is prohibited — Annex II entry 1577 — but ingredients that release it are not, and Commission Regulation (EU) 2022/1181 replaced point 2 of the Annex V preamble: the labelling threshold fell from 0,05 % to 0,001 % and the warning changed from “contains formaldehyde” to “releases formaldehyde”. Any finished product containing a listed substance that releases formaldehyde “shall be labelled with the warning ‘releases formaldehyde’ where the total concentration of formaldehyde released in the finished product exceeds 0,001 % (10 ppm)”.⁷¹⁴ The claims guidance draws the obvious conclusion under truthfulness: “The claim ‘free from Formaldehyde’ is not allowed, if the product contains a formaldehyde releasing ingredient (e.g. Diazolidinyl Urea).”⁹

7. What a “free from” label is worth

There is a published measurement of what the category contains. Tran, Comstock and Reeder queried the Target and Walgreens websites for products marketed as “clean” and analysed the ingredient lists of 1,470 of them. A potential contact allergen was present in 93.8% (1,379 of 1,470); fragrances and botanicals were the commonest at 82.9%, and the single commonest entry of all was the word fragrance, perfume or aroma itself, in 68.2% of the lists.¹⁸ That is a survey of labels rather than of patients, and it measures presence, not reactions — but it is the number, and it is not a small one.

The Green Guides reach the same structural problem from the other end. A free-of claim, they say, may be deceptive even when literally true, in two situations: where the product “contains or uses substances that pose the same or similar environmental risks as the substance that is not present”, or where “the substance has not been associated with the product category”.⁶ The first is substitution. The second is being free of something no product of that kind ever contained.

What a “free from” line reports is one decision a formulator made about one ingredient. It does not report what went in instead, at what concentration, or against which annex ceiling — none of which an ingredient list will give you either, for reasons set out in You Cannot Work Out Percentages From an Ingredient List. And the reasoning underneath it is hazard reasoning, applied one substance at a time and detached from dose, product type and site of use, which is the confusion taken apart in Hazard Is Not Risk. The word on the front of the pack is a shelf label. The document that decides what may lawfully be in the bottle is somewhere else, and it is public.