The European Commission’s proposed Biotech Act creates regulatory sandboxes under EU food law — controlled, time-limited settings in which companies and regulators can jointly work out what data a novel product actually needs before a formal dossier is filed. The proposal then excludes novel foods from those sandboxes by name.

The exclusion is not incidental drafting. According to the position paper published by EuropaBio on 30 March 2026, it sits in Article 56(7) of the proposed Act, which introduces Article 49a(2)(a) into Regulation (EC) No 178/2002, and is justified at Recital 115 on the grounds that certain novel foods may raise “ethical or cultural concerns.”

The practical result is a line drawn in an unusual place. The same fermentation run, the same strain, the same analytical method and the same AI-assisted testing strategy may enter a sandbox when the output is classified as a food enzyme, a food additive, a flavouring or a processing aid — and may not when the output is classified as a novel food. Nothing about the process changes. Only the regulatory class of the thing that comes out of the tank.

For anyone routing a precision-fermented or biomass-fermented protein to the European market, that distinction is the whole story, because essentially every such product is a novel food.

What the proposal actually does

The Biotech Act proposal was adopted by the Commission on 16 December 2025. GFI Europe published its reaction the same day, and links the proposal document as biotech_reg-com2025-1022_act_en.pdf — indicating COM(2025) 1022.

It contains two food-relevant measures pulling in opposite directions.

The measure that helps. The Act expands EFSA’s pre-submission advice. Per GFI Europe, it “proposes that EFSA expand the guidance it provides to companies applying to sell new products, allowing startups to request advice from regulators on the technical and scientific information their applications need to include before making submissions,” and “provides details about additional EFSA staff to ensure that this pre-submission advice function is properly resourced.” EuropaBio describes the same measure as amending the General Food Law “to broaden the scope of EFSA pre-submission advice to study design.”

That is not a small thing. Study design is where novel food dossiers die. It is the difference between a validated application and a file that sits in preliminary validation for a year and is then declared invalid.

The measure that excludes. Alongside it, the Act introduces sandboxes into Regulation (EC) No 178/2002 and carves novel foods out of them.

In scope for a food/feed sandbox Out of scope
Product categories Food enzymes, food additives, flavourings, processing aids, feed, food contact materials Novel foods
Processes and methods Food production processes; biotechnology-driven processes; fermentation techniques; data generation methods; testing strategies; AI- and data-driven tools The same, where the output qualifies as a novel food
Who may establish one Member States only Not applicable: the exclusion operates on product class, not on who sets the sandbox up

Source: EuropaBio position paper, 30 March 2026, characterising the Commission proposal.

EuropaBio states the eligibility rule plainly: those categories may use a sandbox “provided these do not qualify as novel foods.”

There is a second asymmetry in the same proposal, and it has had almost no attention. In the food and feed area, sandboxes can be established only by Member States. In other domains covered by the Act, the Commission may establish them too — EuropaBio points to Recital 168, under which health-sector sandboxes may be set up at EU level by the Commission, potentially with input from the European Medicines Agency. A food sandbox is therefore a national instrument, twenty-seven times over, while a health sandbox can be a single European one.

Sandboxes are also not a shortcut. Under Article 49b, each requires a dedicated “sandbox plan” specifying scope, objectives, duration and safeguards, communicated to the Commission and EFSA. A sandbox prepares a risk assessment; it does not grant market access, and it does not displace Regulation (EU) 2015/2283.

Practical consequence. If you are planning an EU route to market for a fermentation-derived protein, the Biotech Act as proposed changes one thing you can use — richer EFSA pre-submission advice, including on study design — and does not give you access to the instrument designed for first-of-their-kind products. Budget for the dossier, not for the dialogue.

On 27 March 2026, eight food and biotech associations published a joint statement calling the exclusion “not justified on legal, scientific or innovation grounds.” Green Queen reports the statement as carrying eight signatories and names seven of them: FoodDrinkEurope, Food Fermentation Europe, EuropaBio, the European Federation of Associations of Health Products Manufacturers, Food Supplements Europe, the European Agrifood Biotech Alliance and EIT Food. Two of those — the European Agrifood Biotech Alliance and EIT Food — are described by Green Queen as co-funded by the EU itself.

The substance of the objection is narrower and sharper than a general complaint about competitiveness. It is that the stated ground for the exclusion is not a ground EU food law recognises.

EuropaBio’s argument runs through two provisions:

  • Article 7 of Regulation (EU) 2015/2283 permits authorisation of a novel food only where, “on the basis of the scientific evidence available, (it) does not pose a safety risk to human health.”
  • Article 6 of Regulation (EC) No 178/2002 requires risk assessment to be “based exclusively on scientific evidence and carried out in an independent, objective and transparent manner.”

From which: “while ethical or cultural considerations may be relevant at political level, they do not form part of EFSA’s evaluation and are not criteria for authorisation under the Novel Foods Regulation (EU) 2015/2283.”

The signatories then make the point that a sandbox is not an authorisation at all. In their words, it “concerns the scientific and technical preparation of risk assessment, not market authorisation and not broader societal debates,” and including novel foods “would not modify the legal basis for authorisation, nor would it bypass the safety assessment required under Regulation (EU) 2015/2283.”

Their characterisation of the drafting is unusually blunt for a Brussels position paper. Under the heading “Deliberate introduction of incoherence in the adopted proposal,” they write:

“The same biotechnological process, data generation method or AI-supported testing strategy may benefit from a sandbox when applied to enzymes, additives or processing aids, yet be excluded when leading to a product classified as a novel food. This distinction is not based on scientific risk, nor on the nature or complexity of the technology involved, but solely on the regulatory classification of the final product.”

And on proportionality: the exclusion of all novel foods because some may raise ethical concerns is “disproportionate,” since “the vast majority of novel foods authorised under Regulation (EU) 2015/2283 do not trigger such considerations.”

That last claim is worth holding onto, because it is checkable and true in a boring way. The novel food register is dominated by things nobody has ever held a demonstration about — plant extracts, oils, vitamin forms, phospholipid preparations. Cultivated meat is the category that generates the political heat, and it is a small fraction of the pipeline.

Practical consequence. The objection is framed to survive the legislative process, not to win a news cycle. If it succeeds, it will succeed because a co-legislator accepts that Recital 115 imports a non-statutory criterion into an instrument that prepares a science-only assessment. Watch the Parliament and Council amendments to Article 56(7), not the press releases.

Why this matters more for fermentation than for anything else

The categories that keep their sandbox access — enzymes, additives, flavourings, processing aids — are mature. The data requirements are settled. Nobody needs a controlled environment to work out how to characterise a lipase.

The category that loses access is where the data requirements are least settled. EuropaBio makes the point directly: novel foods produced by “precision- and biomass-fermentation, microbial production systems, cell-based techniques” are “often first-of-their-kind and require new types of data, new testing strategies and new assessment approaches. They are precisely the cases where early dialogue with regulators and EFSA on study design, data requirements and methodological approaches would bring the greatest value.”

We have watched that gap produce real outcomes. The Protein Brewery’s Fermotein took six years from filing to EU clearance, as we set out when comparing its EU and US routes. The UK, which is outside this instrument entirely, ran a cultivated-meat sandbox and got two risk assessments out of it — which are not approvals, as we have argued, but are exactly the sort of shared methodological groundwork a sandbox is for.

Green Queen frames the UK programme as “a golden example,” reporting that the sandbox launched in February 2025 with eight companies and produced the first set of safety guidance for novel food approval in the country. We would put it more cautiously — guidance is not throughput, and no UK cultivated product has been authorised — but the direction of the comparison holds. The jurisdiction that left the EU has a novel-food sandbox running. The EU’s proposal excludes novel foods from its own.

Practical consequence. For a company choosing a launch market, this widens an already wide gap. Our five-jurisdiction comparison assumed the EU’s disadvantage was queue length. The Biotech Act, as proposed, leaves queue length untouched for novel foods while shortening it for adjacent categories.

The case for the exclusion

It deserves stating properly rather than being waved away, because there is one.

Sandboxes involve regulators working alongside applicants on the design of the evidence that will later be used to assess those same applicants’ products. That is a real independence question, and it is sharper where a product is contested in public. A Member State authority that spends eighteen months co-designing testing strategies for cultivated meat, in a country where cultivated meat is politically live, has arguably compromised its position before the dossier arrives — not in law, but in perception. The Commission’s recital is a clumsy expression of a defensible instinct: keep the categories that attract political heat out of a collaborative instrument, so that the collaborative instrument survives.

There is also a subsidiarity reading. Novel foods are authorised centrally, for all 27 Member States at once. Sandboxes in food and feed are national. A national sandbox shaping the evidentiary approach to a product that will be authorised EU-wide creates a genuine mismatch — one Member State’s methodological preferences propagating into a decision binding on the other twenty-six. On that reading, the exclusion is not about ethics at all; it is about not letting a national instrument steer a centralised authorisation, and the recital simply gives the wrong reason for the right rule.

The signatories have an answer to the second point — they suggest EFSA could coordinate across Member States, as EMA might in the health sector under Recital 168 — but the proposal as drafted does not do that.

What we could not establish

  • We did not read the Commission proposal itself. Every article number, recital number and quotation of the Act in this piece is taken from the EuropaBio position paper of 30 March 2026 and from Green Queen’s reporting, both of which quote the text directly. We link the Commission PDF as GFI Europe links it, but we did not retrieve it, and we have not independently confirmed that Article 56(7), Article 49a(2)(a), Article 49b, Recital 113, Recital 115 and Recital 168 carry those numbers in the published document. Anyone relying on a specific article number should open the PDF.
  • The eighth signatory. Green Queen reports eight associations and names seven. We could not identify the eighth from the sources we read.
  • The date of Biotech Act II. GFI Europe (December 2025) says a second Biotech Act is “expected to be published in late 2026.” Green Queen (April 2026) says it “is set to be published in Q3 2026” with a focus on agricultural and industrial biotech. Those are not the same window, and we have no Commission source for either.
  • Whether the exclusion has survived. The Act is a Commission proposal. We have not established the current state of Parliament or Council amendments to it, and nothing in this piece should be read as describing enacted law.
  • The WePlanet economic figure. Green Queen reports an analysis that “just €1.4B in investment in alternative proteins could add €111B annually to the bloc’s economy by 2040.” We did not trace that figure to its methodology and are not relying on it.

What to watch

Three testable things.

  1. Whether Article 56(7) is amended. If co-legislators strike the novel-food carve-out, or narrow it to cell-based products only, the eight associations will have won the argument they actually made. If it survives untouched, the “ethical or cultural concerns” ground will have been accepted into an instrument whose stated purpose is scientific preparation.
  2. Whether any Member State establishes a food sandbox at all. The power is national and discretionary. A proposal that creates an instrument nobody uses changes nothing either way. By this time next year there will either be sandbox plans notified to the Commission and EFSA under Article 49b, or there will not.
  3. Whether Biotech Act II names fermentation. WePlanet’s policy brief asks the Commission to list single-cell proteins, mycelium and fermentation-derived ingredients as covered technology categories, and to mandate a fermentation capacity assessment within two years of the Act coming into force. That last item is the one worth wanting: a public inventory of European fermentation capacity would settle arguments this publication currently has to reconstruct from company announcements.

The expanded EFSA pre-submission advice is a real gain and should not be lost in the complaint about the sandbox. But it is advice on how to build a dossier. The sandbox was the mechanism for working out what the dossier should contain when nobody yet knows — and that is the problem first-of-their-kind fermentation products actually have.