US cell-therapy, gene-therapy, and AI-enabled-therapeutic developers approach EU market entry with a mental map of expedited pathways drawn from the FDA suite: Regenerative Medicine Advanced Therapy designation for cell and gene products, Breakthrough Therapy designation where preliminary clinical evidence is strong, Accelerated Approval where a surrogate endpoint will support market entry pending confirmation. The EU Pharma Package, in the Council compromise text of 24 February 2026, introduces a mechanism that none of the FDA programmes map onto. It is not procedural acceleration through a fixed set of authorisation requirements. It is a statutory authorisation to disapply parts of the medicines code, on a product-by-product or category basis, where the inherent characteristics of the product make the standard rules unworkable. The instrument is called a regulatory sandbox. The text remains pre-adoption, with formal adoption expected in autumn 2026 and application of the operative provisions expected in late 2028 or early 2029; article numbering and individual clauses may shift during legal-linguistic finalisation, but the substantive architecture summarised below is stable in the compromise text.
1. A Sandbox in Pharma Law: What the New Mechanism Provides
The sandbox concept entered EU regulatory law from the financial-services tradition (the UK FCA's 2016 pilot is the canonical reference) before being formalised at Union level in Art. 57 ff. of the AI Act. The EU Pharma Package extends the structure into medicines law for the first time. Under Art. 113 of the Regulation, the Commission may, on a case-by-case basis and on a recommendation from the European Medicines Agency (EMA), establish a sandbox where the development and authorisation of an innovative medicinal product or category of products cannot be achieved under the requirements of the Regulation or the revised medicines Directive because of scientific or technical characteristics inherent to the product, and where targeted technical adaptations are considered indispensable.1Council compromise text for a Regulation of the European Parliament and of the Council laying down Union procedures for the authorisation and supervision of medicinal products for human use and establishing rules governing the European Medicines Agency, Council doc ST-6366/26 (24 February 2026), Art. 113. The condition is binary: either the standard rules can accommodate the product or they cannot. The sandbox is not available where they can. Art. 113(10) of the Regulation keeps a separate track open for time-limited pilot projects that test different ways of implementing the existing legislation; those are process experiments, not the product-specific derogations the sandbox provides.
The EU Pharma Package sandbox is not an expedited-review programme; it is a statutory mechanism for disapplying parts of the medicines code where the science is novel enough that the standard requirements cannot be applied to the product at all.
The instrument carrying the derogation is a Commission implementing decision, adopted under the examination procedure, that fixes the sandbox plan: the requirements of the Regulation, of the Directive, and of the Advanced Therapy Medicinal Products Regulation that cannot be complied with; the targeted technical adaptations that substitute for them; the duration of the sandbox; the participants; and the mitigation measures for any residual risk to health or the environment.1Council compromise text ST-6366/26 (n 1), Art. 113(6) and (7). Two gating features are worth flagging at the outset. The first is exclusion of mature programmes: Art. 113(4) of the Regulation directs EMA not to recommend a sandbox for a medicinal product that is already advanced in its development programme. The second is the absence of a developer-initiated entry route. The sandbox is established by Commission decision on EMA recommendation; while the sandbox plan that EMA develops under Art. 113(5) of the Regulation is to be based on data submitted by developers of eligible products, the legislative scheme does not give the developer a right of initiation comparable to the FDA's request-driven designation framework.
The sandbox should not be conflated with the EU flexibility instruments that the medicines code already provides. Conditional marketing authorisation, marketing authorisation under exceptional circumstances, the EMA PRIME priority-medicines scheme, and the earlier adaptive-pathways pilot each adjust the procedural or evidentiary posture of an authorisation within the fixed statutory and regulatory framework. None of them disapplies a provision of the Regulation, the Directive, or the ATMP Regulation. The sandbox is qualitatively different: it is the first Union instrument in medicines law that authorises controlled, time-limited derogation from specified provisions of the medicines code where the science makes those provisions unworkable for a defined product or category. That qualitative difference does not, however, make the sandbox an alternative to those routes at the marketing-authorisation step itself. Art. 114(2) of the Regulation provides that a sandbox-developed product may be placed on the market only when authorised under Art. 13 of the Regulation (standard centralised authorisation), Art. 18 of the Regulation (authorisation in exceptional circumstances), or Art. 19 of the Regulation (conditional marketing authorisation). The sandbox is therefore a derogation layer on top of the three standard procedural routes, not a replacement for them; a sandbox-conditioned conditional MA, or a sandbox-conditioned exceptional-circumstances MA, is a real architectural option rather than a hypothetical one.
2. Scope: Cell, Gene, ATMP, AI-Enabled Therapeutics, and the Boundaries
The scope question is set by Art. 113(1)(a) and (b) of the Regulation, read with its recitals. Art. 113(1)(a) of the Regulation requires that full compliance with the standard requirements be precluded by scientific or regulatory challenges arising from characteristics or methods related to the product, or from scientific or technical characteristics inherent to it. Art. 113(1)(b) of the Regulation requires that those characteristics, on the available scientific evidence, positively and distinctively contribute to the quality, safety, or efficacy of the product or provide a major contribution to patient access to prevention, diagnosis, treatment, or patient care. Recital 133 is more explicit than the operative text about the categories the legislator had in view. It names "innovative technologies, products, services or approaches, at the moment especially in the context of digitalisation or the use of artificial intelligence and machine learning in the life cycle of medicinal products from drug discovery, development to the administration of medicinal products". The recital is non-binding in its category-naming function, but it is the clearest indication available of the products EMA is expected to assess against the gateway in Art. 113(1) of the Regulation.
Advanced Therapy Medicinal Products are addressed through a separate operative route. Reg. (EC) 1394/2007 (the ATMP Regulation) is amended (not repealed) by the Pharma Package and continues to govern cell-therapy, gene-therapy, and tissue-engineered products.2Regulation (EC) No 1394/2007 of the European Parliament and of the Council of 13 November 2007 on advanced therapy medicinal products and amending Directive 2001/83/EC and Regulation (EC) No 726/2004 [2007] OJ L324/121 (ATMP Regulation). Art. 113(2) of the Regulation expressly contemplates targeted derogations to the Regulation, the revised medicines Directive, and the ATMP Regulation. That third limb is the structural opening for ATMPs whose manufacturing, quality, or clinical-evidence architecture does not fit the standard ATMP framework. The sandbox is therefore not a parallel ATMP regime; it is an authorisation to disapply parts of the ATMP framework where the science forces the question. For borderline ATMP architectures that sit awkwardly between the centralised ATMP route and the tightened hospital-exemption regime that the revised medicines Directive introduces in response to documented Member State variability, the sandbox is a third structural option: not a substitute for either pathway, but a vehicle to derogate from specific ATMP provisions where the science requires it.
The AI-enabled-medicinal-product limb sits in a more contested place. A parallel sandbox framework for AI systems generally is set out at Art. 57 ff. of the AI Act.3Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence (AI Act) [2024] OJ L 2024/1689, Art. 57–63. A medicinal product whose lifecycle is materially shaped by AI or machine-learning components, whether in discovery, development, manufacturing, or post-authorisation use, may sit within the eligibility set of both the AI Act sandbox and the Pharma Package sandbox. The relationship between the two regimes is not resolved on the face of either instrument. Art. 113(2) of the Regulation contemplates that targeted derogations apply only to the listed medicines instruments; it does not authorise derogation from the AI Act. The boundary question (which framework's sandbox a developer enters, whether sequential or parallel entry is procedurally possible, how the supervision under each framework coordinates) is one the implementing acts under Art. 115(3) of the Regulation will need to address and which Member State competent authorities will be exposed to before the implementing acts arrive.
The eligibility boundaries are at least as important as the eligibility set. Two are operative. The first is the "not already advanced in development" filter at Art. 113(4) of the Regulation: EMA cannot recommend a sandbox for a product already late in its development programme, which forecloses use of the sandbox as a recovery mechanism for programmes that have run into compliance trouble at the registration stage. The second is the indispensability standard at Art. 113(1)(a) and Art. 113(5) of the Regulation: the adaptations must be indispensable from a technical and scientific viewpoint, not merely useful, and the sandbox plan must so demonstrate. Whether EMA will read indispensability strictly (the standard cannot be applied at all) or pragmatically (the standard can be applied but at disproportionate cost or delay) is a question the implementing acts will not answer and that EMA's first sandbox recommendations will, in practice, settle.
3. EMA and National Competent Authority Roles in Sandbox Supervision
Supervisory authority over a sandbox is split three ways. EMA recommends the sandbox under Art. 113(4) of the Regulation, develops the sandbox plan under Art. 113(5) of the Regulation, levies a fee for its work under Art. 113(5) of the Regulation, and monitors the field of emerging medicinal products under Art. 113(3) of the Regulation; it may engage in preliminary discussions with developers, marketing-authorisation holders, independent experts, healthcare professionals, and patient representatives during that monitoring function. The Commission establishes the sandbox by implementing act under Art. 113(6) of the Regulation, suspends or revokes it under Art. 113(8) of the Regulation, and amends or prolongs it under Art. 113(9) of the Regulation. National competent authorities (NCAs) of the Member States concerned supervise the sandbox in operation under Art. 113(2) of the Regulation: the regulatory sandbox operates under direct NCA supervision for activities that take place on each NCA's territory, and any violation of the conditions in the implementing decision, or any identification of risks to health or to the environment, is notified immediately to the Commission and EMA.
The fee schedule for EMA sandbox activities is not fixed in Art. 113(5) of the Regulation. Whether the fee will be set under the standard EMA fees regulation or under a separately adopted instrument has not been resolved on the face of the text. For the small and medium-sized developers whom Recital 133 specifically contemplates participating in sandboxes, the practical economics of the pathway will turn as much on that question as on the legal architecture surrounding it.
The architecture of Art. 115 of the Regulation adds a layer that is easily overlooked. Art. 115(1) of the Regulation provides that the sandbox does not affect the supervisory and corrective powers of the competent authorities; where risks to public health are identified, NCAs are to take immediate and adequate temporary measures to suspend or restrict use. If mitigation is impossible or proves ineffective, the development and testing process is suspended without delay, and EMA recommends to the Commission that the sandbox be suspended or revoked under Art. 113(8) of the Regulation. The corollary, in Art. 114(5) of the Regulation, is that a marketing authorisation granted under the sandbox is suspended or revoked when the sandbox itself is. This is a different exit dynamic from the FDA expedited-programme suite, where withdrawal of an underlying designation does not vacate the approval the designation supported.
The reporting architecture under Art. 115(4) of the Regulation sits on top of the supervisory layer. EMA, with Member State input, submits annual reports to the Commission, the European Parliament, and the Council, covering the number of sandboxes granted, trends in eligibility, good practices, difficulties encountered, lessons learnt, and recommendations on possible adaptations to the regulatory framework, together with reflections on applying the Regulation and other Union acts within sandbox conditions. The reports and lay summaries are made publicly available. Under Art. 115(5) of the Regulation, the Commission reviews the reports and, where appropriate, brings forward legislative proposals or delegated acts to update the framework. The mechanism is therefore not only a derogation instrument but also a structured legislative-learning instrument: each sandbox is, by design, a data point for the next iteration of the medicines code.
That iteration has a statutory destination. The revised medicines Directive establishes, at Art. 28 of that Directive, an "adapted framework" for categories of products listed in its Annex VII: a permanent counterpart to the time-limited sandbox, populated by Commission delegated act, for products whose inherent characteristics mean the standard requirements cannot adequately assess their quality, safety, or efficacy.4Council compromise text for a Directive of the European Parliament and of the Council on the Union code relating to medicinal products for human use and repealing Directive 2001/83/EC and Directive 2009/35/EC, Council doc ST-6367/26 (24 February 2026) (EU Pharma Package Directive, compromise text), Art. 28 and Annex VII (adapted frameworks); Art. 28(6); Recital 135. Art. 28(6) of that Directive closes the loop, directing the Commission to take the information generated by a regulatory sandbox into account when it adopts those delegated acts, and Recital 135 frames the same progression from experiment to adapted framework. The distinction carries strategic weight for a US developer. A one-off derogation for a single product is a sandbox question; a recurring, class-wide mismatch between a novel modality and the standard rules is an argument for Annex VII listing, which runs on a different timeline, through a different procedure, and without the launch-packaging disclosure that Art. 114(4) of the Regulation attaches to a sandbox-developed product.
4. The Standards That Cannot Be Compromised: Quality, Safety, Efficacy
The Pharma Package is explicit that the derogation power has limits. Art. 113(2) of the Regulation requires that the principles of quality, safety, and efficacy set out by the Regulation and the revised medicines Directive be respected within the sandbox. Recital 133a is more direct: adaptations must ensure that "the principles of quality, safety and efficacy of medicinal products set in the Union legislation are respected". Recital 135 carries the same constraint into the marketing-authorisation step: marketing authorisations granted under a sandbox must ensure those principles. The legislative architecture distinguishes between the technical and procedural requirements that the sandbox plan may adapt and the underlying substantive standards that it may not.
Operationalising that distinction is delicate. Substantive standards are not free-standing in the medicines code; they are embedded in technical requirements (specifications, quality controls, manufacturing standards, clinical-evidence rules) the sandbox is authorised to adapt. A targeted derogation from a specific technical requirement is potentially compatible with continued respect for the underlying quality, safety, or efficacy principle, but the demonstration that it remains so is not self-executing. Art. 113(5) of the Regulation places that demonstration on EMA: the sandbox plan must show that adaptations are indispensable from a technical and scientific viewpoint while the substantive principles are respected, and must propose alternative or mitigation measures where appropriate. The line between adaptation and dilution is therefore drawn in the sandbox plan, in EMA's recommendation, and in the Commission's implementing decision, not in the legislative text.
The marketing-authorisation overlay under Art. 114 of the Regulation maintains the limit. A medicinal product developed in a sandbox may be placed on the market only when authorised under Art. 13, 18, or 19 of the Regulation, and only if the benefit-risk balance is favourable. Initial validity is capped at the sandbox duration set in the implementing act, with renewal possible at the marketing-authorisation holder's request submitted at least nine months before expiry. Derogations from the Regulation and the revised medicines Directive may be carried into the authorisation in duly justified cases under Art. 114(3) of the Regulation: adapted, enhanced, waived, or deferred requirements, each strictly necessary, duly justified, and specified. Crucially, the summary of product characteristics and package leaflet must indicate that the product has been developed as part of a regulatory sandbox, and that condition applies for the sandbox's duration. The labelling rule operationalises a transparency obligation that does not exist in any FDA expedited-programme analogue.
The sandbox derogates from technical requirements, not from the substantive standards of quality, safety, and efficacy; the line is drawn in the sandbox plan rather than in the legislation, and the demonstration that it has been respected is placed on EMA.
5. Comparative: FDA RMAT, Breakthrough, and Accelerated Approval as US Comparators
The FDA pathways that US developers reach for when EU regulatory sandbox enters the conversation are three. Each is structurally different from the EU mechanism, and the gap is the point.
Regenerative Medicine Advanced Therapy (RMAT) designation, codified at 21 U.S.C. § 356(g) and added by section 3033 of the 21st Century Cures Act of 2016, applies to cell therapies, therapeutic tissue-engineering products, human cell and tissue products, and combination products using such therapies or products that are intended to treat a serious or life-threatening condition and supported by preliminary clinical evidence indicating they may address an unmet medical need.521 U.S.C. § 356(g), as added by 21st Century Cures Act, Pub. L. No. 114-255, § 3033, 130 Stat. 1033 (2016) (Regenerative Medicine Advanced Therapy designation). RMAT delivers procedural advantages (intensive FDA guidance, eligibility for priority review and accelerated approval, surrogate-endpoint flexibility) but it operates inside the standard biologics-licensing requirements. It does not authorise the FDA to disapply provisions of the Public Health Service Act or the Federal Food, Drug, and Cosmetic Act that apply to biologics. The accelerative effect is procedural; the regulatory perimeter is not moved.
Breakthrough Therapy designation, codified at 21 U.S.C. § 356(a) and added by section 902 of the Food and Drug Administration Safety and Innovation Act of 2012, applies where preliminary clinical evidence indicates that a drug intended to treat a serious or life-threatening condition may demonstrate substantial improvement over existing therapies on a clinically significant endpoint.621 U.S.C. § 356(a), as added by Food and Drug Administration Safety and Innovation Act, Pub. L. No. 112-144, § 902, 126 Stat. 1085 (2012) (Breakthrough Therapy designation). Breakthrough is largely an intensified-engagement programme: senior FDA staff involvement, rolling review where appropriate, and a tighter feedback loop on development plans. As with RMAT, the substantive requirements applicable to approval do not change.
Accelerated Approval, codified at 21 U.S.C. § 356(c) and operationalised in 21 C.F.R. Part 314 Subpart H for drugs and 21 C.F.R. Part 601 Subpart E for biologics, permits FDA approval for serious-condition products on the basis of a surrogate endpoint reasonably likely to predict clinical benefit, or a clinical endpoint measurable earlier than irreversible morbidity or mortality, with a post-approval confirmatory study requirement.721 U.S.C. § 356(c) (Accelerated Approval), as added by FDASIA, Pub. L. No. 112-144, § 901(a) (2012); 21 C.F.R. pt 314 Subpart H (Accelerated Approval of New Drugs for Serious or Life-Threatening Illnesses); 21 C.F.R. pt 601 Subpart E (Accelerated Approval of Biological Products for Serious or Life-Threatening Illnesses). Accelerated Approval is the closest US analogue to a substantive adjustment of the standard rules: it does not adapt manufacturing or quality requirements, but it permits approval on different evidentiary terms from the default. Even so, the evidentiary alternative is set in the statute and the regulations; it is not negotiated product-by-product as an EU sandbox plan would be.
The structural gap can be put plainly. The FDA programmes accelerate the pathway through the existing requirements, with Accelerated Approval substituting a defined evidentiary alternative for the default endpoint architecture. The EU Pharma Package sandbox authorises the Commission, on EMA recommendation, to disapply parts of the Regulation, the revised medicines Directive, and the ATMP Regulation for a specified product or product category, where the science makes the standard rules unworkable. The FDA suite has no instrument that performs this function. The pre-determined change control plan framework for AI-enabled device software functions, finalised by FDA in December 2024, comes closest in spirit (a structured authorisation to depart from the default fixed-design assumption for AI/ML products) but it operates within the device framework and addresses a specific category of post-authorisation change rather than the development-stage architectural problem the EU sandbox addresses.8FDA, 'Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence-Enabled Device Software Functions: Guidance for Industry and Food and Drug Administration Staff' (Final Guidance, December 2024).
6. Strategic Considerations for US Developers Entering the Sandbox Pathway
The strategic terrain a US developer enters with the sandbox is not the terrain the FDA suite trains for. Several structural questions follow from that, and none of them resolves at a level of generality that releases the developer from product-specific analysis.
The first is timing. The sandbox is foreclosed where the development programme is already advanced. What "already advanced" means in operational terms (does first-in-human position the programme advanced; does completion of pivotal trials; does pre-MAA scientific advice) is unspecified, and EMA's first sandbox recommendations will, in practice, set the operative line. That line is likely to accrete through EMA's existing scientific-advice and Innovation Task Force practice rather than through abstract guidance, and the existing practice in adjacent early-stage contexts is not stable enough to support certainty about where the threshold sits for any specific programme. For a US programme designing its global development plan, the question is whether the EU sandbox option needs to be locked in before the programme reaches a stage at which Art. 113(4) of the Regulation closes it, and whether that consideration changes the timing of EMA engagement under the existing scientific-advice and Innovation Task Force pathways. A US developer who waits until the standard EU regulatory requirements show their edges may find that the sandbox door has already shut.
A related timing layer sits at the clinical-trial-authorisation step. Art. 114(1) of the Regulation directs Member States authorising a clinical trial application for products covered by a regulatory sandbox to take the sandbox plan into consideration. The plan therefore reaches into national clinical-trial authorisations under the Clinical Trials Regulation,9Regulation (EU) No 536/2014 of the European Parliament and of the Council of 16 April 2014 on clinical trials on medicinal products for human use, and repealing Directive 2001/20/EC [2014] OJ L158/1 (CTR), as amended by the EU Pharma Package Regulation (n 1). not only into the eventual marketing authorisation. For a US developer running a multi-Member-State trial programme through the EU clinical trials portal, the practical question is how the timing of the Commission's implementing decision under Art. 113(6) of the Regulation aligns with the trial-authorisation cycle, and how a sandbox plan adopted while trials are ongoing propagates back through the Art. 19 CTR substantial-modification mechanics.
The second is initiation. EMA's recommendation is made on the basis of data submitted by developers of eligible products, and EMA may engage in preliminary discussions during its emerging-products monitoring under Art. 113(3) of the Regulation. But the Regulation does not grant developers a right to initiate, nor does it set a timetable for EMA to respond to a developer-submitted dossier. A US developer cannot apply for a sandbox the way it would apply for RMAT or Breakthrough designation. The procedural levers available to the developer are the existing EMA pathways (scientific advice, the Innovation Task Force, EU-IN Horizon Scanning) repurposed for the sandbox question. Whether and how those existing pathways will be coordinated with sandbox recommendation work under Art. 113(4) of the Regulation is one of the questions the implementing acts under Art. 115(3) of the Regulation will need to settle.
The third is the transparency price. Art. 114(4) of the Regulation requires the summary of product characteristics and the package leaflet to indicate that the product has been developed as part of a regulatory sandbox, for the sandbox's duration. The disclosure is product-facing, prescriber-facing, and patient-facing. The commercial and reputational implications of carrying that label on launch packaging are not addressed in the medicines code and are not analogous to anything in the FDA approval framework, where a Breakthrough or RMAT designation does not become an end-user label. How payers, health-technology-assessment bodies, and prescribers will treat a sandbox-developed product is a question that will be answered in pricing-and-reimbursement and clinical-uptake practice, not in the Regulation.
The fourth is cascade risk. Art. 114(5) of the Regulation ties the marketing authorisation's life to the sandbox's life: if the sandbox is suspended or revoked under Art. 113(8) of the Regulation, the Commission suspends or revokes the marketing authorisation accordingly. The trigger conditions in Art. 113(8) of the Regulation include not only failure to meet the conditions in the implementing decision but also a Commission judgement that suspension is appropriate to protect public health or to avoid serious risks for the environment. The result is that a product launched under a sandbox-conditioned marketing authorisation carries a continuing supervisory exposure that does not lapse on approval. A second, independent revocation route runs through Art. 114(6a) of the Regulation: where EMA concludes that the holder has failed to comply with the specific conditions attached to the sandbox-conditioned authorisation, the Commission may vary, suspend, or revoke it through the Art. 13 procedure of the Regulation, without any sandbox-level suspension having occurred. And the sandbox displaces none of the underlying civil liability: Art. 115(2) of the Regulation keeps every participant, the marketing-authorisation holder included, liable under applicable Union and national law for harm caused by the testing conducted in the sandbox, and obliges them to report to EMA without undue delay anything bearing on the product's quality, safety, or efficacy. Recital 135 spells the exposure out further, naming clinical-trial sponsors, applicants, and any entity involved in the product's lifecycle alongside the holder. How that combined exposure, regulatory and civil, should be allocated in licensing agreements, supply agreements, and clinical-trial-conduct provisions involving sandbox-developed products is not a question with off-the-shelf answers, and US-style boilerplate drafted against an approval-is-final assumption will not accommodate it without modification.
The fifth is exclusivity and protection-term attachment to the sandbox-conditioned MA. The standard regulatory data and market protection regime under Art. 80 of the Directive4Council compromise text ST-6367/26 (n 4), Art. 80 (regulatory data and market protection). attaches to a marketing authorisation by default. Whether a sandbox-conditioned MA, issued under Art. 13, 18, or 19 of the Regulation with derogations specified under Art. 114(3) of the Regulation, carries the standard 8+1 RDP architecture, or whether the sandbox plan may adjust those periods via the "adapted, enhanced, waived or deferred requirements" authority in Art. 114(3) of the Regulation, is not resolved on the face of either instrument. For a US developer modelling EU-exclusivity contribution to pipeline net-present-value, the protection profile depends on what the specific sandbox plan permits and prohibits, and cannot be read off the standard regime.
The supplementary protection certificate regime introduces a parallel uncertainty on the patent side. Under the Supplementary Protection Certificate Regulation,10Regulation (EC) No 469/2009 of the European Parliament and of the Council of 6 May 2009 concerning the supplementary protection certificate for medicinal products [2009] OJ L152/1 (SPC Regulation). the SPC term is calculated from the date of the first marketing authorisation in the Union. A sandbox-conditioned MA with capped initial validity under Art. 114(2) of the Regulation and a nine-months-before-expiry renewal requirement is not a conventional first MA. Whether the SPC clock runs from the sandbox-conditioned authorisation, from a successor non-sandbox authorisation if the sandbox lapses, or under a sandbox plan that itself adapts the SPC interaction, is a question on which the legislation is silent and on which no SPC-grant practice yet exists.
The sixth is interaction with the AI Act sandbox. A medicinal product built on AI Act-regulated components can fall within both that regime's sandbox and the Pharma Package's, and the operative questions are practical: which sandbox to enter, in what sequence or in parallel, and how supervision under the EU AI Office and national AI authorities reconciles with EMA recommendation and NCA supervision. No rule on the face of either instrument coordinates the two.