Who may prohibit a botanical extract, and by what procedure? The challenges to Regulation (EU) 2021/468 turned on the scope of Article 8 and the standard of proof. They left untouched a prior question — the limits of the implementing procedure itself — which bears directly on the draft measure now under scrutiny.
1. From aloin to Annex III: how a prohibition came about
1.1 The origins of the prohibition
Regulatory attention to hydroxyanthracene derivatives (HADs) goes back to Directive 88/388/EEC on flavourings and source materials for their production, whose Annex II set a maximum level of aloin of 0.1 mg/kg in foodstuffs and beverages, with an exception of 50 mg/kg for alcoholic beverages. The legislative technique chosen at the time was a presence threshold: the use of the substance was restricted rather than prohibited.
The shift came with Regulation (EC) No 1334/2008 on flavourings, whose Annex III, Part A, lists aloin among the substances that may not be added as such to food, while Part B sets maximum levels for certain substances naturally present in flavourings and food ingredients with flavouring properties. For aloin, however, no maximum level was set in Part B, in the absence of conclusive toxicological evidence. The Union legislature thus responded to the lack of data by banning deliberate addition and giving up on quantifying natural presence, instead of keeping the threshold that had applied until then. What it did not do, then or later, was prohibit the marketing of food in which aloin is naturally present.
The prohibition of the use in food of certain botanical species containing HADs did not originate in flavouring legislation. It came dressed in the sophisticated garb of a health assessment — one of those ironies in which the wish to endorse a plant's health benefits became the noose that sealed its fate. In 2013, following the assessment of a health claim application submitted by Vivatech through the French competent authority under Article 13(5) of Regulation (EC) No 1924/2006, EFSA's Panel on Dietetic Products, Nutrition and Allergies (NDA) reached two decisive conclusions. First, it confirmed the beneficial effect of HADs on bowel function: a cause-and-effect relationship had been established between the consumption of 10 mg of HADs per day (from plants such as aloe, senna, rhubarb or alder buckthorn) and short-term improvement of bowel function in adults. Second, it attached safety warnings. Relying on the monographs of the World Health Organization and of the EMA's Committee on Herbal Medicinal Products, EFSA introduced safety conditions — in particular, that stimulant laxatives should not be used continuously for more than one to two weeks, with longer use requiring medical supervision, because of the risks of electrolyte imbalance, dependence and impaired bowel function.
The question, then, is what triggered the about-turn that led to a prohibition procedure. The answer lies in the sources of the procedure itself, in particular the Commission working group meeting of 18 February 2016. As reported in the UK Department of Health's information bulletin, after some Member States raised concerns about a potential risk to consumers from the consumption of HADs in food — a debate that ended with the opening of the procedure under Article 8(2) of Regulation (EC) No 1925/2006 — the Commission put two options to the Member States at that meeting: not to authorise the health claim, or to place the substance under Union scrutiny through that procedure. The choice was not without logic. Rejecting a claim backed by a favourable EFSA opinion would have required the Commission to depart expressly from the Authority's opinion and explain the divergence under Article 17(1) of Regulation (EC) No 1924/2006; activating the Article 8 procedure instead meant that the scrutiny itself prevented the claim from being used in the meantime.
Through that channel, on 22 November 2017, EFSA's Panel on Food Additives and Nutrient Sources added to Food (ANS) adopted its opinion on the safety of hydroxyanthracene derivatives for use in food (EFSA Journal 2018;16(1):5090). The Panel concluded that emodin, aloe-emodin and the structurally related substance danthron had shown genotoxicity in vitro; that aloe extracts were also genotoxic in vitro, most likely due at least in part to the presence of those derivatives, although hydroxyanthracene-depleted extracts contained an additional genotoxic component; and that aloe-emodin was genotoxic in vivo in mouse models, whole-leaf aloe extract was carcinogenic in rats, and there was evidence of carcinogenicity of danthron in both rodent species. On that basis, the Panel could not recommend a daily intake of HADs that does not give rise to concern about harmful effects to health, and noted separately that it had no data on the concentrations of these derivatives in the plant parts consumed in a normal diet.
On that footing, the process concluded with the adoption of Regulation (EU) 2021/468 of 18 March 2021, which placed in Part A of Annex III to Regulation (EC) No 1925/2006 aloe-emodin, emodin, danthron and preparations from the leaf of Aloe species containing hydroxyanthracene derivatives, regardless of the amount present; and placed under Union scrutiny, in Part C, preparations from the root or rhizome of Rheum palmatum L., Rheum officinale Baillon and their hybrids, from the leaves or fruits of Cassia senna L., and from the bark of Rhamnus frangula L. and Rhamnus purshiana DC. Two features distinguish this measure from all earlier ones. First, it imposes no maximum level: aloin had been subject to a quantitative limit since 1988, and the NDA Panel's 2013 opinion had itself addressed the risk through conditions of use and safety warnings, whereas Part A of Annex III applies regardless of the amount present. Second, it changes the object of regulation: where the rules had until then targeted the substance, they now target the botanical preparation containing it.
1.2 The actions before the Court of Justice of the European Union
Four actions for annulment were brought against Regulation (EU) 2021/468 before the General Court under Article 263 TFEU: T-189/21, Aloe Vera of Europe; T-271/21, Ortis; T-274/21, Synadiet and Others; and T-302/21, Aboca and Others. The applicants, associations and operators in the food supplements sector, sought the annulment of the prohibitions and the scrutiny measures, pleading the Commission's lack of competence to regulate preparations, breach of the principle of legal certainty because of the vagueness of that term, breach of the principle of proportionality, and manifest error of assessment in imposing an unconditional prohibition without certainty about the real risk at dietary intake levels. The Commission sought dismissal in full, arguing that it had acted in accordance with the precautionary principle and that, genotoxicity of the pure substances having been established and a safe threshold being impossible to set, there was no other public health measure available than to prohibit every extract containing them.
Put that way, the dispute came down to two questions. The first concerned the Commission's powers under Article 8 of Regulation (EC) No 1925/2006: whether the power to act on substances other than vitamins or minerals, or on ingredients containing them, also extends to prohibiting botanical preparations and plant extracts. The second concerned the standard of proof: whether the scientific impossibility of setting a safe daily intake allows a substance to be prohibited unconditionally on health grounds, or whether, on the contrary, the lack of data on actual consumer exposure prevents it from being reliably established that intake greatly exceeds that of a normal diet.
On 13 November 2024, the General Court upheld all four actions and annulled the contested entries, with the exception of the one on danthron, which fell outside the scope of the actions since the applicants did not manufacture products containing it (T-302/21, paragraphs 21 and 22). On the first question, the Commission argued that preparations fall within the scope of Article 8, a reading supported by the definition used in EFSA's 2009 guidance and consistent with the purpose of the provision, since HADs reach food precisely in the form of an extract rather than an isolated molecule. The Court did not accept that interpretation. In the absence of a legal definition, it turned to the usual meaning of the terms and concluded that the concept of preparation is broader than that of substance or ingredient, to the point of being capable of covering the final product, noting that the Commission itself had accepted at the hearing that preparation and ingredient designate different things. As for the guidance, the Court recalled that it is an advisory document with no binding force, and cannot override the meaning of the terms used in the provision conferring a power: where the legislature wrote substance, the Commission cannot read preparation.
On the second question, the Commission's position was that, genotoxicity having been established and no risk-free daily intake being identifiable, the only protective measure available was to prohibit any amount. The Court did not accept that either, and its reasoning deserves to be followed because it is the crux of the case. Article 8 only allows the use of a substance to be prohibited where two cumulative conditions are met: that its addition results in an intake greatly exceeding what is reasonably expected under normal conditions of consumption of a balanced and varied diet, or otherwise represents a potential risk; and that a harmful effect has been identified. The first condition requires a comparison, and every comparison needs two terms: the intake resulting from the product and that of a balanced and varied diet. Without data on the latter, the comparison cannot be made, and it cannot be replaced by an inference that deduces from the absence of data that there is no safe level of intake. The Court therefore held that the absence of a threshold is contrary to Article 8(2)(a)(i), read in conjunction with Article 8(1), of Regulation (EC) No 1925/2006, a formula it repeats in identical terms in all four judgments (T-302/21, paragraphs 83 and 90; T-189/21, paragraphs 59 and 66; T-271/21, paragraphs 87 and 95; T-274/21, paragraphs 78 and 86). But it added a consequence that goes beyond the case: the position the Commission's approach leaves operators in. If recital 20 of Regulation (EC) No 1925/2006 and Article 3(3) of Implementing Regulation (EU) No 307/2012 place on the operator the burden of demonstrating the safety of the ingredient through that very comparison, the same lack of data that prevents the Commission from establishing that the legal conditions are met creates an asymmetry that deprives the operator of any effective means of defence.
Ultimately, what is at issue is not the safety profile of these substances, but the scope of the power to prohibit the ingredients and botanical preparations containing them, the legal terms used to do so, and the standard of proof required.
1.3 Where matters stand
None of these judgments has yet produced legal effects. The Commission appealed in early 2025 (joined Cases C-38/25 P, C-48/25 P and C-55/25 P) and, under the second paragraph of Article 60 of the Statute of the Court of Justice, the annulment of a regulation takes effect only after the period for appeal has expired or, as here, once an appeal has been dismissed. Regulation (EU) 2021/468 therefore remains fully in force and the sales channel for these products remains closed. Yet, far from moderating its approach, the Commission has pressed ahead without waiting for the outcome of the appeal. On 15 July 2026 it sent the Council a new draft regulation (document 11964/26) to move the preparations of Rheum, Cassia and Rhamnus formally from the scrutiny list (Part C) to the prohibition list (Part A). The most contentious aspect is that the Commission has again used the term “preparations”, thereby knowingly running the risk that the Court of Justice will uphold the first-instance judgment and that this new measure, too, will be annulled for the same lack of competence.
2. The limits of the implementing route: essential elements and the scope of the empowerment
All of the above describes a dispute confined to the substantive scope of Article 8 of Regulation (EC) No 1925/2006: what may be included in Annex III and under what standard of proof — in other words, what degree of scientific certainty is required to prohibit it. That approach, though correct, is incomplete. Alongside the debate on the material scope of Article 8 lies a prior problem of a different legal nature, which the four judgments do not address because the original actions did not raise it: the procedure through which, and the form in which, that power must be exercised.
The contested measures are implementing acts adopted under the regulatory procedure with scrutiny provided for in Article 14(3) of the basic Regulation — a mechanism governed by Article 5a of Decision 1999/468/EC (the Comitology Decision) and reserved for measures of general scope designed to amend non-essential elements of a basic act. In principle, an implementing measure may neither alter the essential elements of the enabling act nor extend the scope of the powers conferred. If including botanical preparations in Annex III exceeds the objective scope of Article 8, as follows from the General Court's judgment, the consequence is not only a breach of the enabling provision: the measure falls outside what that procedure can decide at all. Far from being speculative, this perspective shifts the axis of the debate. It directly affects the validity of the draft regulation now under way and provides an autonomous ground of challenge, wholly independent of the outcome of the appeal.
The significance of that procedure deserves attention, because it is not a mere formality but a control mechanism with legal consequences of its own. Under Article 5a(3) of Decision 1999/468/EC, once the competent committee — the Standing Committee on Plants, Animals, Food and Feed, as the July 2026 draft itself designates it — has delivered a favourable opinion, the Commission must submit the draft measures without delay to the European Parliament and the Council, which have three months to oppose their adoption. That opposition is not discretionary; it must be based on one of the following grounds, each of which is in itself a sufficient basis for blocking the measure: that the draft exceeds the implementing powers provided for in the basic act; that it is not compatible with the aim or content of the basic act; or that it does not respect the principles of subsidiarity or proportionality. If the Parliament, by a majority of its component members, or the Council, by qualified majority, opposes the draft on such grounds, the Commission may not adopt it. The first ground of opposition and the first reason why the General Court annulled Regulation (EU) 2021/468 are, it should be noted, exactly the same: exceeding the powers conferred.
That control is not hypothetical; it was actually exercised. The draft that became Regulation (EU) 2021/468 was transmitted on 1 December 2020 and submitted to scrutiny by the European Parliament and the Council under Article 5a(3)(a) of Decision 1999/468/EC. Shortly afterwards, on 13 January 2021, the Council formalised its decision not to oppose the adoption of the measure (document 5221/21). Three years and ten months later, the General Court held that the very same provision exceeded the power conferred by Article 8 of Regulation (EC) No 1925/2006, by regulating botanical preparations where the parent provision only allows action on substances. The comitology control thus failed in its essential function of safeguarding the limits of competence: neither the Parliament nor the Council noticed the Commission's overreach, a defect of legality that forced the sector to bear the effects of a measure vitiated, according to the first-instance judgment, by lack of competence until the General Court stepped in.
The reason why this procedure carries limits that cannot be crossed lies in settled case law of the Court of Justice. The adoption of the essential rules in a field is reserved to the Union legislature and cannot be delegated; implementing measures therefore cannot amend the core elements of the basic act or supplement it with new essential elements. And — decisively for the thesis defended here — the classification of an element as “essential” is not left to the institution adopting the measure, nor to the legislature's discretion, but must be based on objective factors amenable to judicial review, having regard to the characteristics and particular features of the field concerned. So held the Court of Justice, sitting as the Grand Chamber, in its judgment of 5 September 2012, Parliament v Council (C-355/10), which annulled a provision adopted under the regulatory procedure with scrutiny precisely because it encroached on matters reserved to the legislature.
Applied to Article 8 of Regulation (EC) No 1925/2006, the question comes into sharp focus. That Annex III may be amended by implementing measures is not in dispute: the provision itself says so, and that is the whole point of the empowerment. What is open to question is whether the transfer of power also covers determining what kinds of things may appear in that annex. Deciding, within the categories set by the legislature — substance, and ingredient containing a substance — which ones are prohibited, restricted or placed under scrutiny is one thing. Widening those categories by adding a third, the “botanical preparation”, which the legislature did not provide for and whose scope, according to the General Court, may extend to the final food product, is quite another. The former is a legitimate exercise of the delegated power; the latter redefines its perimeter de facto. And the perimeter of an empowerment can hardly be regarded as a non-essential element of the act conferring it, since it is what sets the exact boundary of the delegated power and marks where the reserve of legislation begins.
The most serious objection to all this is obvious and should be put in its strongest form. Since Article 8 expressly provides for Annex III to be amended by implementing measures, the legislature clearly intended the content of that annex to be decided outside the ordinary legislative procedure; arguing that determining what appears there affects essential elements could therefore empty the empowerment of content. Moreover, the 2006 legislature did not define “substance” or “ingredient”, which suggests that it avoided setting closed categories. On that premise, reading the provision so as to exclude botanical extracts would deprive it of effect in the very situation its recital 20 contemplates: intake through extracts and concentrates.
Strong as it is, the objection does not reach the core of the argument. Delegating the specification of the annex's content is one thing; delegating the determination of the categories of what may appear in it is another. That the legislature did not define “substance” and “ingredient” does not mean it did not use them as delimiting terms; it simply means their scope must be established by the usual methods of interpretation — which is exactly what the General Court did in concluding that a “preparation” fits neither concept. Finally, the effectiveness argument actually proves something other than what it claims. If Article 8 is insufficient to reach botanical extracts, the conclusion is not that it should be read expansively, but that the instrument needs direct intervention by the legislature. That an empowerment has fallen short is no legitimate basis for widening it.
It is not argued here that the General Court formally decided the competence question now raised, since the original actions did not put it directly. But the way the Court reasoned shows that the competence problem was present in its analysis. The signs are clear: in denying interpretative value to the EFSA guidance for lack of binding force — a strict argument about the sources admissible for delimiting a power — in requiring a uniform interpretation of the legislature's concepts, and in stressing that the Commission itself accepted at the hearing the distinction between preparation and ingredient, the General Court traced the path. That methodology shows that the Court did not merely review a scientific assessment; it reviewed the scope of a conferral of power.
3. The draft under way and the mechanics of Part C
All this would be of purely doctrinal interest if the debate were closed. It is not. On 15 July 2026, the Commission sent the Council a new draft regulation (document 11964/26) to move the preparations of Rheum, Cassia and Rhamnus formally from the scrutiny list (Part C) to the prohibition list (Part A) of Annex III. The draft, based on Article 8(5) of Regulation (EC) No 1925/2006, already has the favourable opinion of the Standing Committee (SCoPAFF) and provides for a twelve-month transitional period. The critical point is that the Commission persists in using the concept of “preparations”. This linguistic and conceptual defect is no longer confined to the route used to adopt Regulation (EU) 2021/468; it is replicated, in the same wording as the measure annulled at first instance, in this new procedure.
The point matters because inclusion in Part C is not a neutral state. The bar for placing a substance under scrutiny (Part C) is markedly lower than that for prohibiting it (Part A): the mere possibility of a risk and the existence of scientific uncertainty suffice. Once in Part C, the system starts a countdown: operators have a strict twenty-four-month period to submit safety studies, and the Commission must decide within four years. At that point the burden of proof is reversed entirely. To enter the scrutiny list, it is enough for the Commission to voice suspicion; to leave it, the operator must demonstrate safety.
That this machinery leads to automatic prohibition is not a hypothesis but an observed reality. The precedent of Yohimbe (Pausinystalia yohimbe) is highly instructive. Having entered Part C on mere suspicion, the failure of a fragmented sector to submit evidence was enough for the Commission to prohibit it outright in 2019, without any new scientific indication of toxicity. Applied to plants with a long tradition of consumption, for which no one has gathered costly exposure data because it was never required, this procedural architecture makes prohibition by default the system's natural outcome.
The resulting situation deserves to be stated with strict objectivity. The July 2026 draft regulation is subject to scrutiny by the European Parliament and the Council through the same procedure that was followed for Regulation (EU) 2021/468. As explained, one of the exhaustive grounds for opposition is precisely that the proposal exceeds the implementing powers provided for in the basic act. There is, moreover, a General Court judgment holding that Article 8 of Regulation (EC) No 1925/2006 does not allow botanical preparations to be regulated. Even so, the new draft uses that term again, now relying on paragraph 5, whose wording likewise refers to “substances”. It is not for this analysis to predict whether the Parliament or the Council will exercise their veto; institutional practice in this field has traditionally been not to oppose, as the precedent of document 5221/21 shows. It is for this analysis to point out that the ground for opposition is available, that its basis is set out in a judicial decision, and that the period for invoking it is short.
The question, however, goes beyond this file. If the thesis defended here is correct, the real problem is not the unfortunate wording of two successive regulations, but an unbridgeable gap between the legal instrument and the reality it seeks to regulate. Article 8 was designed to act on isolated chemical substances and defined ingredients, but botanicals enter the food chain as complex extracts. As long as that conceptual gap persists, every measure the Commission adopts on a preparation will be exposed to the same charge of illegality. To avoid it, the Commission will have to identify the molecule responsible and quantify its exact presence — precisely what the precautionary route of the scrutiny list allowed it to skip. This is not a mere drafting defect: it shows that the legal empowerment has fallen short of the regulatory ambition demanded of it.
4. The outlook for botanical extracts
This terminological conflict did not begin in 2021. In 2019, Regulation (EU) 2019/650 had already placed Yohimbe and its preparations in Part A, using the same technique that the General Court has now held to fall outside Article 8. The contrast with later measures is equally instructive. Monacolins from red yeast rice were placed in Part A worded strictly as a “substance”, not a “preparation”; and the restrictions on green tea extracts rich in EGCG initially took the form of quantitative limits of use in the restrictions list (Part B). There is, then, a legal route that does withstand the General Court's objections: identifying the responsible substance precisely and setting its conditions of use. The price of that route is that it requires analytical characterisation and exposure data that the convenient precautionary route of the scrutiny list allowed to be skipped.
Three scenarios are therefore possible, and none is neutral. If the Court of Justice, on appeal, adopts a broad reading of “ingredient containing a substance”, the Commission will regain its effectiveness against extracts and concentrates at the cost of sacrificing the textual limit of the law; the whole debate will then move to the standard of proof and proportionality, questions still undecided. If the Court of Justice upholds the first-instance judgment, the Commission will have to reformulate its prohibitions in terms of “substance”, specifying which derivative it bans and at what concentration — which would, paradoxically, produce far more precise regulation, with greater safeguards than the current one. Finally, if the solution were legislative, the natural vehicle would be to align Regulation (EC) No 1925/2006 with the Lisbon Treaty rules (Articles 290 and 291 TFEU), which would require reopening Article 8 and allow its vocabulary to be corrected at the root.
This structural friction is aggravated, finally, by a regulatory absence that is not an oversight but a documented political choice. Article 4(8) of Directive 2002/46/EC required the Commission to present a report before July 2007 on the advisability of establishing specific rules, including positive lists, for plants and other substances. The Commission's report (COM(2008) 824) concluded that such European harmonisation was not justified. As a result, the use of these ingredients remains fragmented under the rules of each Member State, forcing the sector to navigate between divergent national catalogues such as the BELFRIT list (Belgium, France, Italy), the German Stoffliste or the Austrian list of plants, whose criteria do not coincide. The resulting asymmetry deserves to be stated plainly: Union law harmonises with surgical efficiency what is prohibited, through Article 8, and leaves to national law what is permitted. A provision originally conceived as a mere safeguard clause thus carries, dysfunctionally, the entire harmonising burden of a sector that lacks a common positive regime.
Javier Gómez Sanz is the founder and director of Food Tech Legal. Comments on this analysis may be sent to editorial@foodtechlegal.com. Enquiries about a specific product or case are handled through the quote form. The original Spanish version of this analysis is available here.