Food allergen law is written by source. Annex II of Regulation (EU) No 1169/2011 lists, at entry 7, “Milk and products thereof (including lactose),” and at entry 3, “Eggs and products thereof.” The trigger is where the substance came from.
Precision fermentation breaks that structure, because the protein does not come from anywhere. A beta-casein assembled by a genetically modified yeast is, molecule for molecule, the thing a cow makes. It has never been near a cow.
Three regulators have now had to decide what goes on the label, and they have reached three different answers. The Australian and New Zealand regulator’s answer is the most revealing, because it says out loud that its own answer is legally uncomfortable and takes it anyway.
What FSANZ proposed, and what it conceded
Application A1342 seeks approval for a fermentation-derived beta-casein protein preparation, referred to throughout as BCPP1, produced by a genetically modified strain of the yeast Komagataella phaffii. The applicant, per the FSANZ-hosted executive summary, is Eden Brew Pty Ltd. Supporting Document 2 — the labelling assessment, dated 9 July 2026 — sets out what FSANZ proposes to require.
The conclusion:
“FSANZ considers that the required name ‘milk’ should be declared when BCPP1 is present in a food for sale or is sold to a caterer or manufacturer, because this would best support the safety and informed choice of consumers with milk allergy.”
Then, three paragraphs earlier, the concession:
“The beta-casein in BCPP1 is derived from yeast and technically would not meet the definition of ‘milk’ in the Code (see section 2.1 of this report). This means a ‘milk’ declaration could be viewed as misleading.”
FSANZ resolves this by ranking its statutory objectives rather than by reinterpreting the definition:
“FSANZ considers the potential for a milk declaration about BCPP1 to be misleading is outweighed by these safety concerns, and therefore its presence in food should be declared with the required name ‘milk’ next to the name of the ingredient and in the summary statement.”
Mechanically, this would be done by amending the table to section S9—3 of the Code to add BCPP1 as a food for which the required name ‘milk’ must be declared. It is a product-specific listing, not a general rule — which matters, because it means the next fermentation-derived allergen needs its own amendment.
FSANZ goes further than a back-of-pack declaration. It also proposes:
“a front-of-pack mandatory advisory statement indicating that the product is not suitable for people with milk allergy. The advisory statement would apply to dairy analogue products containing BCPP1, and to products containing BCPP1 that are represented in a form which expressly or by implication suggests that the food does not contain milk.”
That second limb is the operative one for anyone marketing on an animal-free platform. A product whose presentation implies it does not contain milk triggers the front-of-pack statement.
Practical consequence. If A1342 proceeds as drafted, an ANZ label for a BCPP1 product needs three things: ‘milk’ next to the ingredient name, ‘milk’ in the summary statement, and a front-of-pack line saying it is not suitable for people with milk allergy. Artwork built for a US or UK launch will not carry the third.
The regulator-versus-regulator disagreement
FSANZ considered simply relying on existing allergen rules — the approach the UK and EU take — and rejected it in terms that are unusually direct for a regulatory document:
“While it would align with intended regulatory approaches in jurisdictions such as the United Kingdom and the European Union, which rely on general allergen labelling requirements, FSANZ considers these approaches do not adequately address the identified risk.”
Set that against what the UK Food Standards Agency published seven weeks earlier. In a post dated 16 June 2026 by Lavanya Kala, Head of Innovation Policy, the FSA writes:
“Existing allergen labelling rules apply fully to precision fermented food products, just as they do for conventional products.”
The FSA identifies exactly the same hazard FSANZ does, and in almost the same words:
“Because BLG is a potent cow’s milk allergen, these products may not be suitable for consumers with a milk allergy, something that may not be immediately obvious from how they are marketed.”
“Terms like ‘animal free’ can be conflated by consumers with ‘plant-based’ or even ‘dairy free’, which could put people with allergies at risk.”
Same risk, same diagnosis, different remedy. The FSA’s remedy is that businesses should apply existing rules carefully — “we strongly encourage businesses to take a cautious approach.” FSANZ’s remedy is a new mandatory front-of-pack statement.
It is worth being precise about the status of the FSA post, because it is easy to over-read. It is a blog post in the “Business guidance” category. It cites no statute, no regulation and no statutory instrument by name; it links only to FSA guidance pages. It announces no new requirement. Its own framing is advisory. Anyone treating it as a statement of UK legal position is treating it as more than it says it is.
The FSA does, however, make one point neither of the others makes, and it is the most practically useful sentence in any of these documents:
“If you are developing a product for the UK market, do not assume that labelling requirements authorised elsewhere will automatically meet UK requirements.”
The US route: a doctrine, applied consistently
The United States settled this earliest and has been the most consistent. The mechanism is the FD&C Act’s definition of a major food allergen, and the FDA applies it the same way in every response letter we read. From the letter closing GRN 863 in March 2020:
“The FD&C Act defines a ‘major food allergen’ as one of eight foods or food groups… or a food ingredient that contains protein derived from one of those foods. [β-]lactoglobulin produced by fermentation requires labeling under the FD&C Act because it contains protein derived from milk.”
The doctrinal move — how a protein that never touched milk is nonetheless “derived from” it — appears in the FDA’s allergen guidance, Edition 5, January 2025:
“We consider these proteins to be ‘derived from’ major food allergens because they are produced in a manner that uses the major food allergen’s DNA sequence.”
That is a genuinely different theory from FSANZ’s. FSANZ concedes the product is not milk and overrides the point on safety grounds. The FDA holds that it is derived from milk, because the sequence is. One is a policy trade-off; the other is a definitional claim. Both produce a milk declaration, but they would diverge on any product where the sequence is engineered away from the natural one.
The FDA guidance also floats a phrasing the other jurisdictions have not adopted:
“FDA recognizes, however, that additional information may be helpful in some cases to make the relationship to the source ingredient clear, such as ‘Contains milk-derived protein,’ and we are available for consultation.”
The guidance is expressly non-binding — it carries the standard “Contains Nonbinding Recommendations” marking — and FSANZ notes the same.
The US clearances, assembled
| GRN | Notifier | Substance | Production organism | FDA letter signed |
|---|---|---|---|---|
| 863 | Perfect Day, Inc. | β-lactoglobulin | Trichoderma reesei QM6a-PD1 | 25 March 2020 |
| 1056 | Remilk Ltd. | β-lactoglobulin | Komagataella phaffii yRMK-66 | 15 February 2023 |
| 1145 | Imagindairy Ltd. | β-lactoglobulin | Aspergillus oryzae Ao_st0002 | 18 December 2023 |
| 1200 | Vivici B.V. | β-lactoglobulin | Komagataella phaffii VIPLA | 28 February 2025 |
| 1219 | TurtleTree | Recombinant bovine lactoferrin isolate | Komagataella phaffii M020 | 7 May 2025 |
| 1241 | Bon Vivant SAS | β-lactoglobulin | Aspergillus oryzae | 16 September 2025 |
| 1247 | Shanghai Changing Biotechnology Co., Ltd. | β-lactoglobulin | Kluyveromyces lactis CCTCC M20241460 | 19 September 2025 |
| 1249 | Onego Bio, Inc. | Ovalbumin preparation (egg) | Trichoderma reesei TR112, gene from Gallus gallus | 16 September 2025 |
Every one of those letters carries an allergen labelling section requiring declaration of milk — or, for GRN 1249, egg. The wording is near-identical across notices, with one variation worth noting: the lactoferrin letter (GRN 1219) says the isolate requires labelling “because it is a milk protein,” rather than because it contains protein derived from milk.
There is a small discrepancy in the record here, and it goes the direction that matters. FSANZ’s own survey of the US position states that there are “at least seven” cleared proteins and lists “five Beta-lactoglobulin whey proteins (GRAS notices 863, 1145, 1200, 1241, 1247).” That list omits GRN 1056 — Remilk’s β-lactoglobulin, which the FDA closed in February 2023 and which is plainly the same class of substance. There are at least six.
FSANZ’s “at least” makes the statement defensible rather than wrong, and nothing in its conclusion turns on the count. But a formal regulatory document surveying another jurisdiction is the sort of thing that gets cited onward as a complete enumeration, and this one is not.
Practical consequence. If you are benchmarking your dossier against the US precedent set, work from the FDA’s own inventory rather than any secondary list, including a regulator’s. We link the inventory.
The EU, which has not said
This is the gap in the record, and we want to be plain that it is a gap rather than an answer.
Annex II of Regulation (EU) No 1169/2011 is headed “SUBSTANCES OR PRODUCTS CAUSING ALLERGIES OR INTOLERANCES.” Despite that heading, the entries themselves are drafted by source, not by substance:
“3. Eggs and products thereof;”
“7. Milk and products thereof (including lactose), except: (a) whey used for making alcoholic distillates including ethyl alcohol of agricultural origin; (b) lactitol;”
Even the carve-outs are derivation-based. The general footnote reads: “And the products thereof, in so far as the process that they have undergone is not likely to increase the level of allergenicity assessed by the Authority for the relevant product from which they originated.”
Annex II says nothing about a protein produced without an animal. It neither includes nor excludes one. Anyone asserting that Annex II settles this question is reading in what is not there.
We were unable to find a statement from any EU institution — Commission, EFSA or a Member State authority — on how Annex II applies to a fermentation-derived milk protein. The only characterisation of the EU position we could source is FSANZ’s, which notes that no fermentation-derived allergenic protein has completed EU novel food approval, that applications must include an allergenicity assessment, and that if such products contain known allergens then the general labelling requirements would apply. That is a second-hand account, and we present it as one.
The absence is not surprising. As we have set out before, the EU has cleared no precision- fermented protein, so no European label has yet had to exist. The question becomes live the day the first one does.
What a formulator actually has to do
| United States | United Kingdom | Australia / New Zealand (proposed) | |
|---|---|---|---|
| Declare milk in the ingredient list | Yes | Yes, under existing rules | Yes, as the required name ‘milk’ |
| Summary/“contains” statement | Yes | Yes, under existing rules | Yes |
| Front-of-pack advisory | Not specified | Not specified | Yes, mandatory for dairy analogues and for products implying no milk |
| ‘Animal-free’ / ‘vegan’ claims | Not addressed in the letters read | Flagged as a consumer-confusion risk; not prohibited | Considered and not proposed for prohibition |
| Basis of the requirement | Protein “derived from” milk via the DNA sequence | General allergen rules apply “fully” | Safety objective overrides a definitional mismatch |
Two further naming constraints sit underneath that table and are easy to miss.
First, generic dairy ingredient names do not travel. FSANZ states that “the existing generic ingredient names ‘milk protein’ and ‘milk solids’ would not apply to BCPP1 because their use depends on the definition of ‘milk’ in the Code,” and that “‘cheese’ as a generic ingredient name would not apply to a cheese analogue made with BCPP1.” So the product must be declared as milk for allergen purposes while being barred from calling itself a milk ingredient. Both at once.
Second, the FDA has explicitly declined to bless a name. The GRN 863 letter states that its use of the term β-lactoglobulin “is not our recommendation of that term as an appropriate common or usual name,” and notes that the office assessing safety did not consult the office responsible for nutrition and food labelling on the point.
Third — and this one is a trap for anyone naming a product around its plant base — FSANZ warns that “names indicating a plant-based protein source (for example, ‘oat milk’) may lead consumers with milk allergy to assume the product is suitable for them, even where it contains BCPP1.”
Practical consequence. Three label variants, not one with regional overprints. And the descriptor you use in marketing is now a regulatory trigger in ANZ, not just a positioning choice.
The case against FSANZ’s front-of-pack statement
FSANZ’s own document supplies the strongest objection to it.
The proposal rests on the premise that consumers misread ‘animal-free’ as safe. FSANZ concedes the evidence for that premise is thin. On consumer comprehension it records that “no evidence was found on consumers’ ability to accurately identify allergens in this context,” and on the specific descriptor: “There was no direct evidence that consumers would interpret ‘animal-free’ representations in the same way [as ‘vegan’].”
So a mandatory front-of-pack statement — the most intrusive labelling intervention short of a prohibition — is proposed on an inferred risk rather than a measured one. A regulator that required demonstrated consumer harm before mandating front-of-pack text would have chosen the FSA’s position, and would not obviously have been wrong.
There is a second objection. FSANZ’s declaration is product-specific, added to a table by name. That is administratively tidy for one application and completely unscalable across a pipeline that already has eight cleared proteins in the US alone. Every new fermentation-derived allergen needs its own amendment, each taking its own consultation cycle. A general rule — of the kind the FDA reached by interpretation — handles the ninth product without doing any more work.
Against that: industry itself has been telling regulators the descriptor is a problem. FSANZ records that a joint industry guide concluded ‘animal-free’ “is unlikely to be suitable for expected regulatory labelling requirements in APAC,” while separate consumer research found ‘animal-free’ “was found to be highly effective for consumer communication” — and that the same research “did not evaluate how terms like ‘animal-free’ interact with allergen expectations.” The industry’s preferred term is effective at communicating, untested on the one dimension where being misunderstood is dangerous.
What we could not establish
- Any EU-institution statement on whether fermentation-derived milk or egg proteins fall within Annex II. Not found. The EU characterisation in this piece is FSANZ’s, not the EU’s.
- Annex II from EUR-Lex directly. Every EUR-Lex URL we tried returned a JavaScript bot challenge rather than the text. The wording and entry numbers here are verified from legislation.gov.uk, which states that its text derives from EUR-Lex under Commission Decision 2011/833/EU, and which we checked in both the as-adopted and current revised versions. They are identical for entries 3 and 7.
- The beta-casein content of BCPP1. FSANZ SD2 says the preparation is “approximately 5.5% beta-casein.” The applicant’s own executive summary, hosted on the same FSANZ site, describes it as “approximately 25% A2 beta-casein and 75% co-purified host cell proteins.” We cannot reconcile those and have not tried to.
- The current status of A1342. SD2 is a Call for Submissions supporting document. We did not establish whether the consultation has closed or whether the draft variation has been approved. The FSANZ page that would confirm this renders only with JavaScript.
- The legal status of the FSA post. It cites no instrument. We do not know whether it reflects a settled FSA legal position or an early statement of intent.
- Whether any product containing a precision-fermented milk or egg protein is on sale in the UK or EU. FSANZ states none has market authorisation in either. We did not verify that independently.
- Canada and China. Everything FSANZ says about them we read only in FSANZ’s account. We did not fetch Health Canada’s Letter of No Objection or the Chinese standard, and neither should be cited from this piece as primary.
- Perfect Day’s partner labelling requirements, which FSANZ describes citing a 2021 company source. We did not check whether they still apply.
What to watch
- Whether the A1342 front-of-pack statement survives consultation. If it does, Australia and New Zealand will have the strictest precision-fermentation allergen regime in the world, and the first one that regulates the marketing claim rather than only the ingredient list.
- Whether FSANZ generalises. A product-by-product listing in S9—3 will not survive contact with a pipeline of eight-plus cleared proteins. Watch for a horizontal amendment covering fermentation-derived allergenic proteins as a class.
- The first EU novel food authorisation of a fermentation-derived dairy protein. That is the moment the EU has to say something about Annex II, and whatever it says will bind 27 markets at once. Our reading of the text is that it can go either way on the drafting alone.
- Whether anyone tests the premise. All three regulators are acting on an assumption about how consumers read ‘animal-free’. FSANZ says plainly that the evidence does not exist. One properly designed study would settle whether the front-of-pack statement is necessary or merely prudent — and would be cheaper than the compliance cost of getting it wrong in either direction.