A Review of Product Safety Regulations in the European Union
AA Review of Product Safety Regulations in the European Union
Jukka Ruohonen a, ∗ a Department of Computing, University of Turku, FI-20014 Turun yliopisto, Finland
Abstract
Product safety has been a concern in Europe ever since the early 1960s. Despite the long and relatively stable historicallineage of product safety regulations, new technologies, changes in the world economy, and other major transformationshave in recent years brought product safety again to the forefront of policy debates. As reforms are also underway, there isa motivation to review the complex safety policy framework in the European Union (EU). Thus, building on deliberativepolicy analysis and interpretative literature review, this paper reviews the safety policy for non-food consumer productsin the EU. The review covers the historical background and the main legislations, administration and enforcement of thepolicy, standardization and harmonization, legislations enacted for specific products, notifications delivered by nationalsafety authorities to the EU’s alert system, recalls of dangerous consumer products, and the liability of these. Based onthe review and analysis of these themes and the associated literature, the current policy challenges are further discussed.
Keywords: safety, consumer protection, harmonization, standards, liability, literature review, European Union, GPSD
1. Introduction
Safety has reached a global priority in the face of theglobal COVID-19 crisis. Yet safety—understood in thepresent context as a risk to human health—has longbeen on the agenda of legislators around the world, in-cluding those in the European Union. Although globalpandemics—from the Spanish flu through the swine flu tothe present crisis—have often captured the attention inpopular discourse, law-backed preparations for hazardousaccidents have been implemented from the 1960s onward.These preparations address also concerns that range fromterrorism, crime, and radicalization to traffic, pollution,and environmental hazards, all of which may also includesafety consequences [5, 35]. Interestingly, many of the inci-dents that originally prompted the preparations have beenforgotten or buried to history books. The production,storage, and transport of chemicals is a good example:the global and European safety legislations for chemicalsmoved forward through crises; dioxin in 1976, toxic oilsyndrome in 1981, methyl isocyante in 1984, amerithraxin 2001, and so forth and so on [14]. Consumer productsdo not cause such large-scale accidents, but the safety con-sequences from these often affect more humans than indi-vidual hazardous accidents. Regulation of safety require-ments for consumer products is also particularly difficult.Product safety is a subset in the larger jurisprudenceof consumer protection. Throughout the world, the ratio-nale builds upon the economic and information asymme-tries between producers and consumers; the latter are in ∗ Corresponding author.
Email address: [email protected] (Jukka Ruohonen) a weaker position with respect to the former, both withrespect to bargaining power and knowledge [63]. As thehistory of the automobile industry vividly demonstrates,the incentives of producers to blindly pursue profits haveresulted in many dangerously defective or even hazardousproducts that have put consumers’ lives at risk [59]. Giventhat later on safety became an important competitive ben-efit in the car industry [53], black and white perceptionsshould be avoided, as always. Nevertheless, the generalrationale of product safety regulations is to protect theweaker party. A particular focus has historically beenplaced upon vulnerable groups, such as children, disabled,and elderly [46]. This protection rationale appeared onthe legislative agenda already in the early 1960s both inEurope and the United States [34]. It did not take long forit to appear also in the European Economic Community;the early safety directives were passed in the early 1970s.However, fragmentation has prevailed to the presentday. Although perhaps not as visibly as in some other ar-eas of consumer protection, also European product safetylegislations have suffered from fragmentation and incoher-ence. A partial explanation originates from the domain;it is difficult to legislate consumer goods and services dueto the pace of innovation and technological progress. Butanother partial explanation stems from different culturesand historical trajectories; the power struggles betweenthe organized interests of consumers and producers af-fected regulatory traditions differently in different memberstates [1]. A similar struggle between organized interestshas affected consumer law at the EU-level [49]. A furtherpartial explanation can be found among producers; therehave often been diverging or even strictly competing inter-ests between sectors and producers, and their locations in
Preprint submitted to Elsevier February 9, 2021 a r X i v : . [ c s . C Y ] F e b ifferent member states and regions [13]. Likewise, therehave been occasional arguments that the EU’s safety reg-ulations are used for protectionist objectives [34], partic-ularly against Chinese products [30]. Concerns such asenvironmental consequences have intensified the struggles.These challenges translate into research challenges; theEU’s product safety policy has always been notoriouslycomplex and difficult to understand. This complexity pro-vides a motivation for the present short review on the keylegislations in Europe. But there is a further motivation:reforms are already underway for European product safetylegislations due to the rapid technological progress and itsimpact upon the economy and consumers with it. Elec-tronic commerce, data, platforms, and other ingredientsof contemporary economy have also changed the incen-tives, externalities, and asymmetries between producersand consumers [31, 33]. Although product safety doesnot perhaps weigh as much as security and privacy inthese new circumstances, it is not difficult to imagine alsonew safety risks arising from, say, artificial intelligence,robotics, and cyber-physical systems, along many othertechnological development trends. Indeed, many novel butwretched safety incidents have already occurred and beencataloged for artificial intelligence applications [42]. Ofthese incidents, the legacy of Elaine Herzberg is perhapsthe most memorable and saddest example; she was thefirst pedestrian having been killed by a self-driving car.
2. Approach
A few preliminary remarks are in order before the ac-tual review: a few basic concepts should be clarified, somewords should be said about the approach to the review,and something should be noted about the empirical dataused alongside the review for a few illustrative points.
Safety, security, hazard, and risk—among many relatedconcepts—are domain-specific, debated, and generally am-biguous terms [5]. In terms of information security, whichis a subset of a larger security conundrum, a risk is some-times understood as a probability that an attack occurs;other times it is more specifically seen as a conditionalprobability resulting from a threat and a vulnerability.From this perspective, protection of (information) secu-rity implies protection against intentional attacks, whereassafety is more about unintentional lapses [35]. Given thecontext of consumer products, it is also useful to frame un-intentional harms to those that have consequences for thehealth and well-being of humans. This framing aligns withthe concept of hazard, which is often understood merelyas a potential source of harm [53]. Despite the differences,for illustrative purposes, the basic information securityconcepts can be translated to the safety context: a vul-nerability could be a defect in a product that exposes asafety threat to human health, such as, say, a suffocation, a strangulation, or a serious electromagnetic disturbance.In this review, as well as in the EU regulations, such safetythreats of consumer products exclude social, psychological,and related factors with potential health consequences. Itis also important to underline that the particular regu-lations considered exclude food products, medicine anddrugs, and occupational health risks, among other things.Product safety is presumably the earliest case of arisk-based approach to regulation in the EU. Since the1990s safety regulations have relied on a precaution prin-ciple: dangers to health and environment should be takeninto account through systematic, scientifically based riskanalysis [58]. Although definitions vary across domains,a risk-based approach according to the European safetyregulations is seen to generally cover three dimensions:risk assessment, risk management, and risk communica-tion [51]. These vary from a product to another. For manyproducts—from chemicals to cosmetics, risk assessmentsmay involve rigorous laboratory testing. For some otherproducts, including software products, assessments rangefrom the following of standards, documentation, and soundengineering practices to quality controls and safety verifi-cation [8, 45, 50]. Risk management, likewise, varies acrossproducts. For many products, including both tear-and-wear hardware products and software products, life-cyclemanagement is usually present; a product should be safethroughout its intended life in the hands of consumers [53].It should be also stressed that risk management is not onlystrictly about safety; for producers cost-benefit analysis isoften present as well [34, 58]. In the European Union riskcommunication carries a particular weight: whenever arisk is found from a product, producers should communi-cate the risk to public authorities and consumers. Withthese clarifications of the basic terminology in hand, theapproach taken for the review can be briefly elaborated.
The review approach taken follows the tradition ofpractice-oriented policy analysis. Unfortunately—just likewith safety, there are no commonly agreed definitions forpolicy analysis. Roughly, policy analysis revolves aroundthe questions of what, how, and why governments do whatthey do, and what difference does it make [15]. In thisshort review the focus is on what they do with a policy.This policy refers to a set of European legislations andstandards designed to ensure the safety of non-food con-sumer products. Safety in itself is seen as the primary (butnot necessarily the only) answer to the why question.Then, it seems reasonable to maintain that most practi-tioners of policy analysis would agree that it: (a) requiressensitivity to a given policy space, which is neither lim-ited to a particular polity nor a decision-making system;(b) cannot be separated from politics; and (c) involves apractical motivation of careful evaluation of problems and,whenever possible, different solutions to these. The policyspace for product safety is not limited to the EU’s par-liamentary decision-making; national safety administra-2ions and standardization organizations—among others—possess considerable power in both shaping and interpret-ing the overall safety policy. Such power leads to politics,which, in the present context, is also shaped by politiciansas well as the organized interests of producers and con-sumers. In what follows, only a limited focus is placed onagency and the intentions of political actors. The identi-fication of problems and bottlenecks satisfies the practicalmotivation, although policy recommendations are kept toa minimum as reforms are already underway in the EU.Also the epistemological bases for policy analysis vary.At least interpretative, narrative, normative, critical, his-torical, positivist, evidence-based, and deliberative policyanalysis frequently appear in the literature. The last onesuits the purposes of this review well. Formulated in theearly 2000s as a critical response to the distinctively pos-itivist policy studies at the time, deliberative policy anal-ysis builds on three pillars: interpretation, deliberation,and practice-orientation [27, 39]. Although a few descrip-tive statistics are presented, the analysis is based on aqualitative interpretation of the main policy artifacts, thesafety legislations in the EU. In terms of deliberation andpractice-orientation, the intention is to cover the main ar-guments in the historical and present policy debates, as-sessing the relative merits of these and giving a unique in-put through a synthesis. At the same time, the approachtaken is a literature review. Quantitative reviews (suchas meta-analysis) and protocol-based (such as systematicliterature reviews) approaches do not go well with policyanalysis. The reasons for this claim are many, ranging froma need to assess historical developments to the requirementto cover both politics and policies. Thus, a traditional, in-terpretative approach is followed with the literature; thegoal is to develop a comprehensive understanding and crit-ical assessment of existing knowledge via in-depth read-ing [4]. This classical approach does not mean that the col-lection of literature would have been unsystematic. All rel-evant databases were queried, among them ScienceDirect,HeinOnline, Taylor-Francis Online, SpringerLink, WileyOnline Library, IEEE Xplore, and SAGE Journals.
The few descriptive statistics presented are basedon the EU’s Community Rapid Information System(RAPEX) [23]. Established in the early 2000s, RAPEXis a database for tracking notifications sent by nationalsafety authorities about dangerous consumer products, ex-cluding food and pharmaceutical products but includingclothing, cosmetics, toys, electronic appliances, and manyother product types. It is administrated by the Euro-pean Commission like many analogous safety tracking sys-tems, including those related medicines, drugs, food prod-ucts, serious cross-border health threats, and chemicalsincidents [14, 47]. Although a database was establishedalready in the 1990s for alerts on dangerous consumergoods [57], it is no longer publicly available online; RAPEX provides records from 2005 onward. In total, n = 28 , F r equen cy Figure 1: Annual Notifications (excluding 2021)
As can be concluded from Fig. 1, annual submissionamounts have been relatively stable from the 2010s on-ward. After accelerating growth in the 2000s, roughlyabout two thousand entries were filed to RAPIX each year.It is difficult to interpret these magnitudes, but given thesize of the EU’s internal market and the amount of con-sumer products circulating within and across it, a coupleof thousand dangerous products per year seems modest.
3. Review
The short review covers seven distinct but overlappingthemes: the general, historical background, and the coreregulations, administration, standardization, product cat-egories, notifications, recalls, and liability, respectively.
Product safety is present in the Treaty on the Function-ing of the European (TFEU). Like in many policy domains,the legal basis builds upon the functioning of the internalmarket for the free functioning of goods, people, services,and capital. In addition to this general clause specified inthe TFEU’s Article 26, the member states have agreedupon the prohibition of import and export restrictionsamong themselves with Articles 34 and 35. However, thesubsequent Article 36 states that indiscriminating restric-tions are possible for goods on the grounds of protectingthe health and life of humans, animals, or plants, amongother things. Furthermore, Article 12 states a general goalof consumer protection when legislating and enforcing lawsin the European Union, and Article 191 extends the over-all health protection goal toward environmental consider-ations. Before the Lisbon Treaty in 2007, which incorpo-rated the Maastricht Treaty as the TFEU, similar protec-tive clauses were specified in the Treaty Establishing theEuropean Community. As the European Economic Com-munity was turning to the European Union, tighter har-monization was required also for product and food safety.Instead of focusing on specific products, common Eu-ropean legislative framework was sought on four strate-gic areas: fair trading, public health, public controls, and3onsumer information, unified by standardization [9, 32].These strategic goals were based on the so-called New Ap-proach to regulation, which has generally been perceivedas a success and thus an important factor for the Euro-pean integration. In terms of jurisprudence, it started asa response to a seminal 1985 case in the Court of Justice ofthe European Union (CJEU). The decision reached by thecourt was significant for two reasons. Both were related tothe internal market. On one hand, it established the so-called principle of mutual recognition (i.e., goods sold inone member state must have access to the whole internalmarket); on the other, it mandated a set of essential pub-lic interest safety requirements for products [61]. The NewApproach led to Directive 92/59/EEC on general productsafety. It is based on five general principles, as follows:1. The directive’s scope covers products intended forconsumers, including new, used, or reconditionedproducts but excluding second-hand products. Safeproducts, in turn, refer to products that under nor-mal, or reasonably foreseeable conditions of use, poseonly a minimal risk. The antonym is a dangerousproduct. When assessing a risk, packing, instructions,and related factors should be taken into account inaddition to a given product’s characteristics in itself.2. The general safety requirements obligate producersto place only safe products on the internal market.These are specified either by European standards or,in the absence of such standards, national laws en-acted in the member states. In addition to thesesafety requirements, producers must provide adequateinformation to consumers and ensure that identifica-tion of individual products and product patches ispossible after their release to the internal market.3. The member states are obliged to ensure compli-ance through properly authorized national authori-ties . Their obligations range from compliance mon-itoring and safety checks to ex ante prohibitions formarket entry and ex post withdrawal of products.4. The member states are further mandated to provide notifications to the European Commission about anymeasures taken regarding dangerous products.5. The
Commission is empowered to inform other mem-ber states in case a given member state undertakes anemergency action for dangerous products. If a EU-wide solution is required, the Commission has also aright to enforce a withdrawal of a dangerous product.Finally, common EU institutions coordinate productsafety issues between the member states.The Maastricht treaty prompted an update to the prod-uct safety directive. In particular, the TFEU’s Article 169strengthened the legal basis for consumer protection, in-cluding on health and safety issues. The resulting policy-making in the 1990s led to Directive 2001/95/EC, alsoknown as the general product safety directive (GPSD). Itis the directive in force today. Although the directive mademany amendments and clarifications to the 1990s one, the five general principles remained largely unchanged.Among the amendments and clarifications are obligationsfor supply chain distributors of products. The GPSD alsosubstantially extended the notification framework and in-formation exchange procedures with the RAPEX archi-tecture. Despite the architecture, further alterations wererequired for more efficient monitoring of the internal mar-ket. To this end, the 1990s Regulation (EEC) 339/93 wasrepealed with Regulation 768/2008 (hereafter, MSR) formore rigorous market surveillance of dangerous products.In general, the MSR strengthens the GPSD. A particularemphasis is placed on national accreditation authorities,serious risks, and further information exchange provisions.Today, the GPSD and the MSR are the main effectivelegislations. That said, reforms were attempted through-out the 2010s, largely due to the emergence of electroniccommerce. Already in 2013 a new regulatory package wasattempted, but it got struck in a legislative limbo, as didthe results from a 2016 evaluation by the Commission [61].In practice, only the mutual recognition principle has beenclarified with Regulation 764/2008 and its successor, Reg-ulation (EU) 2019/515. Both had only a minimal im-pact upon the existing product safety legislations. A newconsultation for product safety was launched in 2020 ina conjunction with larger planned reforms on electroniccommerce, digitalization, and related aspects affecting thesingle market. Even though it is too early to evaluatethe impact from the feedback, it is worth remarking thatthe political trenches were dug as could be expected. Re-garding electronic commerce, platform companies, suchas eBay [16], argued that platforms should be exemptedfrom tighter constraints, while consumer and civil societygroups, such as BEUC [60], pointed out that a substantialamount of products purchased via platforms were alreadynon-compliant with the EU laws and technical standards.
The administrative framework is typical to the Euro-pean Union. Excluding some sectors (such as food andmedical devices) in which European safety agencies exist,administration is typically decentralized to the memberstates. The EU-level is generally reserved for coordinationand information exchanges. In fact, none of the articles inthe TFEU establish a particular requirement for a EU-levelcompetency. Instead, the legal basis for pan-European ad-ministration of product safety has largely been justifiedwith the harmonization measures specified in the TFEU’sArticle 114 [28, 51]. In practice, these measures includestandardization and information exchanges. These werespecified also in the GPSD and the MSR alongside the en-forcement at the national level (Articles 6–10 in the formerand Articles 2 and 16 in the latter). As the administrationis typical to the EU, so are the impediments and problems.To some degree, decentralized administration at the na-tional level has maintained the historical cross-country dif-ferences and thus fragmentation across Europe. Depend-ing on a study, three, four, or five different consumer pro-4 ssential safety requirementsVoluntary use ofEuropean standards Technical detailsspecified in law Risk-based safety assessment procedureAccreditation via conformity assessmentSelf-assessmentby a producer Authorizedthird-party audits Legal complianceGeneral Product Safety DirectiveProduct-specificlegislation Product-specificlegislation
Figure 2: The General Logic of the Safety Policy in the European Union (adopted from [19] with alterations) tection regimes can be identified in Europe. Accordingto one classification, there has been a Nordic negotiationmodel (industry associations and individual companies ne-gotiate directly with consumer associations for commonpolicy goals), a protection model with France as an exam-ple (consumer associations and the state have had a stronginfluence upon policy goals), and an information model with Austria and Germany as examples (industry associa-tions and the state have had a dominant role) [1]. In addi-tion, it is possible to identify further models by casting thefocus on the British-influenced administrative tradition aswell as the eastern and southern member states.These regimes are not unique to product safety adminis-tration and consumer protection in general. For instance,a comparable administrative system—including the noti-fication mechanism—has been established for Europeancyber security [55]. Also privacy and data protection inthe EU share many similarities in terms of administrationand coordination—as well as the associated problems [54].These problems include a lack of resources, funding, andexpertise in some countries, generally inconsistent enforce-ment, poor testing facilities in some countries, powerless-ness in terms of sanctions, diverging legal interpretations,and general fragmentation [29, 37, 61, 62]. Given the Com-mission’s limited power—it cannot even act as an arbiterin case the member states disagree on product safety is-sues, many of the problems likely prevail in the foreseeablefuture. All this said, there has been a high degree of coher-ence for some consumer products due to standardization.
The EU legislations for product safety rely stronglystandardization. According to the New Approach, legisla-tive harmonization establishes essential requirements re-quired for an entry to the internal market, but discretionis allowed for producers regarding the technical standardsthat fulfill these requirements [12]. The role of standardsin the safety policy’s general logic is illustrated in Fig. 2.The difficulties for covering new products through regu-lation explain the difference between legislation and stan-dardization. But the difference can be interpreted also as a way to separate politics from technical expertise;or, rather, to balance parliamentary legislation procedureswith industry self-regulation [13, 34]. This balancing hasalso raised questions about the dominance of the latterover accountability [29]. In terms of legal wordplays, ac-cording to the GPSD, a product is deemed safe wheneverit complies with a given European or national legislation.Yet, according to Articles 3 and 4 in the GPSD, a productis presumed safe insofar as it conforms to voluntary na-tional standards transporting harmonized European stan-dards. Here, European standards refer to those draftedby European standardization bodies in accordance withDirective (EU) 2015/153. It includes technical specifi-cations that specify quality, performance, testing, safety,packaging, and related dimensions, but excludes radio andtelevision broadcasting, telecommunications, and financialservices, among other things. The major European stan-dardization bodies for harmonized standards are the Eu-ropean Committee for Standardization (CEN), the Eu-ropean Telecommunication Standards Institute (ETSI),and the European Committee for Electrotechnical Stan-dardization (CENELEC). All three are linked to interna-tional standardization organizations and their committees.From the 2000s onward, European standardization effortshave generally leaned toward industry consortia and co-regulation in order to improve competitiveness and inno-vation [3, 30]. Safety standardization is not an exception.In some sectors, there has been some confusion betweencompliance to European standards and compliance to Eu-ropean safety legislations [40]. In many sectors, how-ever, compliance questions are relatively straightforwardfor producers due to the work done by the European stan-dardization organizations who translate the essential re-quirements into technical specifications. The CEN, in par-ticular, has released thousands of technical standards overthe decades. With these standards, accreditation providesthe presumed safety conformity requirement [13]. Theprime example is the CE marking established with Di-rective 93/68/EEC, which was later augmented with theCommission’s Decision 768/2008/EC. It is not a certifica-tion but a declaration of conformance and due diligence.5ypically, the CE marking is obtained by self-certificationand following of harmonized European standards, with orwithout additional assessments by a national safety agencyor third-party auditors [12, 50]. However, the CE markingaccreditation has been an exception rather than the rule.Fragmentation is present also in terms of standardiza-tion. Many complex products need to comply with multi-ple legislations and multiple standards. A further problemhas been the GPSD’s generality; Article 3 implies that thepresumed safety assumption rests on national standard-ization, which is required to transpose the various Euro-pean standards. It is thus no wonder that the three Euro-pean standardization organizations have called for unifiedEU-level standards, which, according to their position, areachievable via better coordination, funding, and strategicthinking [20, 22]. Finally, it is important to underline thatcompliance with European standards and their nationaltranspositions is voluntary. While complying with theseprovides the presumed safety, it is still possible to placeproducts that are only deemed safe to the internal market.If the products turn out to be unsafe, withdrawal of thesemay follow by national authorities, as soon discussed.
The New Approach pushed the regulatory work towardgeneral product categories and essential safety require-ments for these. A few notable legislations for productcategories are enumerated in Table 1. These are the direc-tives that were originally amendment with the CE-markingDirective 93/68/EEC. Of these, the old voltage Directive72/23/EEC is particularly noteworthy. It was this direc-tive after which the New Approach was largely modeled.Even with this small snapshot of the specific EU legis-lations, it can be concluded that the scope is wide; frommachinery and vehicles to medical devices and toys. Fur-thermore, many products (such as machinery or motor ve-hicles) must comply with multiple legislations [50]. It isalso worth remarking that some notable consumer prod-ucts, such as cosmetics [48], are missing from the listing.A further important point is that the relation of the spe-cific legislations to the GPSD and its predecessor has beena source of some confusion [30]. In principle, the GPSDand associated national laws should apply in case there areno specific legislations for a given product category. TheMSR is also explicitly specified as lex spesialis ; it is appli-cable only insofar as there are no specific legislations formarket surveillance, as is the case with drug precursors,medical products, vehicles, and aviation. As the GPSDalso states that the risks specified in it are applicable un-less overruled by a specific legislation, it should be takeninto account alongside any existing specific legislations.To get a sense on the actual, realized safety risks acrossthe categories, Fig. 3 shows the ten most frequent prod-uct categories (alongside a catch-call group for other cat-egories) across all entries filed to the RAPEX, from 2005to mid-January 2021. Toys, clothing and textiles, motorvehicles, and electronic devices constitute the majority of reported unsafe consumer products. Together these cate-gories account for about 68% of all RAPEX filings. Par-ticularly the large amount of dangerous toy products isinteresting and surprising. It is, however, difficult to spec-ulate the reason for this result; one explanation could bethat the safety of toys is vigorously enforced by the na-tional safety authorities. This would align with the longtradition of considering children as a particularly vulner-able consumer group [46]. Numerous European standardshave also been specified for toys [40]. In addition to toys,many filings were made about clothing, textiles, and fash-ion items, motor vehicles, and electrical appliances.
Other categoriesProtective equipmentLighting chainsChemical productsLighting equipmentChildcare, children's equipmentCosmeticsElectrical appliances and equipmentMotor vehiclesClothing, textiles and fashionToys 0 2000 6000Frequency
Figure 3: Top-10 Product Categories
Three additional points are worth making about theRAPEX entries. First: besides potential variance in termsof enforcement practices, these observations likely reflectthe large amount of consumer products in falling to thesecategories. Second: the product categories further reflectthe typical risk types shown in Fig. 4. For instance, about34% and 32% of dangerous toy products had caused chem-ical and chocking injuries, respectively. From all productswith injury risks, about 41% were toys, 16% cosmetics,and 16% clothing, textiles, and fashion items. Last: thefrequencies in Fig. 3 correlate with the origin from whichthe products were imported to the internal market. Ascan be seen from Fig. 5, the clear majority of the danger-ous products were imported from China. For instance, asmuch as 78% of the dangerous toy products were manufac-tured in China. As such, the observation cannot be strictlyinterpreted to imply that Chinese products would be par-ticularly risky, given that most toy products are manufac-tured in China. Nevertheless—given that about 19% oftotal imports to Europe were from China in 2020 [24], itseems sensible to relate the observation to the increasinglycomplex supply chains for consumer and other products.6 able 1: Notable Safety Directives for Specific Products
Sector Earliest legislation Latest legislationBurning gaseous fuels Directive 90/396/EEC Directive 2009/142/ECConstruction products Directive 89/106/EEC Regulation No 305/2011Electrical equipment (voltages) Directive 72/23/EEC Directive 2014/35/EUElectromagnetic compatibility Directive 89/336/EEC Directive 2004/108/ECHot-water boilers (fueling) Directive 92/42/EEC –Implantable medical devices Directive 90/385/EEC –Machinery Directive 89/392/EEC Directive 2006/42/ECNon-automatic weighing instruments Directive 90/384/EEC Directive 2014/31/EUPersonal protective equipment Directive 89/686/EEC Regulation (EU) 2016/425Simple pressure vehicles Directive 87/404/EEC Directive 2014/29/EUTelecommunications terminal equipment Directive 91/263/EEC Directive 93/68/EECToys Directive 88/378/EEC Directive 2009/48/EC
OtherElectromagnetic disturbanceEntrapmentDrowningAsphyxiationCutsHealth risk / otherSuffocationDamage to hearingMicrobiologicalEnvironmentDamage to sightFireBurnsStrangulationElectric shockChokingChemicalInjuries 0 2000 4000 6000 8000Frequency
Figure 4: Risk Types
The notification mandates set in Directive 92/59/EECwere modeled after existing procedures laid down for someother sectors. Notably, notification procedures for phar-maceuticals were laid down already in the mid-1970s andearly 1980s with Directives 75/319/EEC and 81/851/EEC.These were followed by Directives 82/894/EEC and89/662/EEC concerning animal diseases and products ofanimal origin, respectively. Furthermore, a year after theChernobyl disaster, Council’s Decision 87/600/Euratomhad established a system for rapid exchange of informa-tion in radiological emergencies. The GPSD generalizedthese procedures toward a general requirement: accordingto Article 5, both producers and distributors must inform
Other countriesIndiaPolandJapanUnited KingdomItalyFranceTurkeyUnited StatesGermanyChina 0 5000 10000 15000Frequency
Figure 5: Top-10 Countries of Origins a national authority whenever they know that a productcontains safety risks, and coordinate with them on any pre-ventive measures taken. In practice, the mandate largelyrests on producers’ own risk assessment and managementprocedures [30]. With accreditation, such as CE marking,however, the national accreditation authorities should in-form each others according to the MSR’s Article 12. It isalso worth mentioning notification requirements betweenthe national authorities and the Commission, as well asinformation exchanges in situations requiring rapid inter-ventions, as specified in the GPSD’s Article 11. Finally,there is the notification requirement toward the subjectsbeing protected, the consumers; as specified in Article 16,information about safety risks should be available to thepublic according to transparency and other good adminis-tration practices. Some criticism has been leveraged aboutpassivity in this informing obligation [28]. At the EU-levelthis criticism finds its target in the RAPEX infrastructure,7s well as in its use by national authorities and media.
Product withdrawals were an important element alreadyin the legacy Directive 92/59/EEC. Both the GPSD andthe MSR clarified the powers granted to national authori-ties in this regard. The MSR’s Article 21 and the GPSD’sArticle 5 stipulate that national authorities have the rightto ban new market entries as well as withdraw existingdangerous products from the market. This withdrawal de-terrence has provided an important incentive for producersto ensure safety of their products [10]. Though, Article 5in the GPSD somewhat loosens the obligations for produc-ers by emphasizing that recalls should only be used as alast resort. This concession is understandable because re-calls have been a controversial issue for both producers andregulators: for the former—besides plain economic losses,these may interfere with insurance schemes; for the latter,there has been a fear that producers will externalize dif-ficult recall decisions to them [30]. In terms of productsafety, however, withdrawals are particularly important asthese concern products consumers are already using.
Other countriesPolandGreeceFinlandCyprusBulgariaFranceUnited KingdomSpainHungaryGermany 0 2000 4000 6000 8000Frequency
Figure 6: Top-10 Country Filings to RAPEX
Recalls have also been frequent according to the empir-ical RAPEX sample. By using simple regular expressionsearches, about 35% and 33% of the products’ descrip-tions contain the words recall and withdraw , respectively.Other keywords are far less common; these include ban (13%), reject (6%), correct (5%), destruct (3%), among afew other auxiliary terms. Therefore, it can be concludedthat the national safety authorities have also actively usedtheir product recall powers. This observation reflects themany individual high-profile withdrawal cases with seriouseconomic implications [53]. But the RAPEX does not, un-fortunately, contain enough information to deduce whether the many product withdrawals undertaken were done aftera proactive notification from a producer or a distributor.Nor is it sufficient to deduce about the criticism that somecountries have preferred voluntary corrective actions [62].If filings to the RAPEX is used as a proxy, a large cross-country variance is indeed present. For instance, morefilings were made from Hungary than from France and theUnited Kingdom (see Fig. 6). This observation reinforcesthe earlier points about general fragmentation and inco-herence. A further relevant question is whether the rela-tively low share of market entry bans indicates improve-ment potential for more proactive ex ante enforcement.
A final important aspect of the New Approach was theadoption of the product liability Directive 85/374/EEC.It extends the hybrid approach of risk-based assessments,harmonization and standardization, and market surveil-lance with explicit allowance of litigation. Safety is ex-plicitly spelled in the directive’s Article 6, and legal basisfor penalties is possible by translating the “polluter paysprinciple” implicitly present in the TFEU’s Article 191 tothe safety context [51]. The directive introduces strict lia-bility for producers, but there are several means by whicha producer may argue against a claimant: improper useof the product in question, the presence of a defect at thetime when the product was launched or marketed, the lackof scientific knowledge at the time when the defect was no-ticed, the presence of the defect despite compliance withlegislations and standards, and so forth [52]. For thesereasons, Directive 85/374/EEC can be interpreted to be astrict liability law only on paper—in practice, the neces-sity to demonstrate the damage, the defect, and the causalrelation between them make the directive negligence lia-bility law instead [25, 64]. It has nevertheless provided anincentive for producers to ensure safety [53]. All in all,it is important to stress that a “regulation through liti-gation” approach has generally been limited in the safetydomain; reforms have been pushed forward by the orga-nized interests of producers, consumers, and experts ratherthan by litigation, which is historically paradoxical due tothe CJEU’s central role behind the New Approach [51].Although hundreds of cases have been brought to courtsthroughout Europe [17], high-profile cases affecting largeconsumer groups have been rare for consumer goods.There is no single explanation why the litigation-centricapproach did not fully emerge. In addition to lobbying,the reasons include: the experiences in the United States,where strict liability was established for product safety inthe 1960s but which came under criticism in the 1980sfor its alleged economic impacts, the state-of-the-art de-fense provisions in Directive 85/374/EEC, the Europeantradition of compensations through welfare states, under-development of insurance schemes for producers, a lackof contractual relationships in some cases, litigation costsfor consumers, and the recall deterrence, among other8hings [6, 10, 17]. History offers another plausible explana-tion: Directive 85/374/EEC was enacted before there wasany talk about the subsidiarity principle, before the Maas-tricht Treaty [25]. Regarding more recent times, an alter-native interpretation would be that the GPSD, the auxil-iary product-specific legislations, and European standardshave worked efficiently in preventing dangerous goods fromentering the internal markets to begin with. However, notall products are protected from safety threats equally. Al-though the question has been more about security thansafety—a distinction, which, as noted, is not always clear,software liability has not generally progressed despite along global debate [45, 56]. As soon discussed, softwareand information technology products and services havealso introduced other major security and safety challenges.
4. Discussion
This review addressed consumer product safety regu-lations in the European Union. Although the topic iscomplex—covering tens if not hundreds of legislations,many of which address highly specialized products andscience associated with these, some general points canbe briefly made about the safety regulations. These arelargely based on a hybrid approach. Thus, first, theGPSD and the MSR provide the umbrella legislations aug-mented with product-specific legislations and Europeanstandards. Both are relevant for compliance. Second, how-ever, the hybrid policy is generally based on the precau-tion principle. Risk analysis carries a particular weight.Third, the administration is decentralized to the memberstates. Like in many related policy domains, the EU-levelis mainly reserved for coordination and information ex-changes. Fourth, even though possible, litigation has beenrelatively rare; product recalls remain the main deterrence.Although recommending change for the sake of changeis one of the deadly sins of policy analysis [43], there aremany indicators of a reform need. Many of these have alsobeen acknowledged in the EU. The Commission’s recentsummary report [18] of the feedback on the GPSD sufficeto outline the main factors behind the need. These are: • Although the Commission pointed out inconsistency only in terms of food-imitating products, incoherenceand fragmentation are a larger problem for the wholesafety policy. As is common in the EU, the reason ispartially explained by both vertical (national policiesversus EU policies) and horizontal (variance acrossthe member states) incoherence. The issue goes be-yond the safety legislations. For instance, the productliability directive has not been uniformly adopted inEurope [17]. Another notable aspect relates to Euro-pean standards, which still need to be transposed tonational standards. The position of European stan-dardization organization seems rather unequivocal:there is a need for unified standards at the EU-level. Incoherence and fragmentation are not the only rea-sons; industry competitiveness is also a concern withstandards. According to interviews [8], in some sec-tors particularly small and medium-sided enterpriseshave faced challenges in selecting appropriate stan-dards and implementing these for their products. • Enforcement has been a closely related problem. Al-though the situation cannot be described as a race tothe bottom whereby some member states would delib-erately weaken their enforcement responsibilities [25],resources, expertise, legal interpretations, and incen-tives, among other factors, vary across the memberstates. As the Commission noted, the effectivenessof recalls is a particular concern because many con-sumers continue to use dangerous products despiterecall notices and withdrawals of product stocks insale. As for potential solutions, as the European Par-liament emphasized in its 2019 resolution [21], furtherharmonization, sufficient resourcing, and better coor-dination may improve the situation. As there are EUagencies for safety in some sectors, it is also worthcontemplating whether or not a move toward EU-leveladministration might offer a longer term solution. • Market surveillance has been a third common prob-lem. Closely related to the enforcement problems, ac-cording to the Commission, there has been a lack oftools and instruments to impose effective sanctions.As with security, the complex global supply chainsconstitute a major problem for product safety in Eu-rope and elsewhere. Traceability, product integrity,and supply chain management are increasingly im-portant for safety risk management and efficient re-calls [41]. Analogous points apply to software prod-ucts and their security [11]. Although consumer prod-ucts may be an exception, it can be generally arguedthat European approaches are not sufficient alone; themanagement of large-scale hazards requires global so-lutions. Furthermore, within Europe, the enactmentof the MSR has led to uneven requirements for dif-ferent products; the incoherence across products andacross Europe applies also to post-market activities. • Online platforms are an increasing problem for prod-uct safety and its enforcement. As such, the issueis hardly new. The difficulty to identify sellers andhold them accountable on electronic commerce mar-ketplaces was recognized already in the early 2000s [6].Yet, only recently have the European safety author-ities cast their attention to the issue. As an exam-ple: in total, from 2018 onward, 244 submission tothe RAPEX have mentioned either eBay, Amazon, orboth. Although the amount is tiny compared to theamount of consumer products delivered via these plat-forms, it is still sufficient to conclude that the problemis already recognized also on the side of enforcement.A particular problem with these new platforms re-9ates to their business model; they act as intermedi-aries for third-party sellers whose responsibilities—oreven identities—are not always clear. Like with sup-ply chains, the issue is also largely global, and, un-fortunately, existing international arrangements aresoft guidelines at best in this regard [2]. As con-sumers increasingly purchase products from sellers lo-cated outside of the EU, ensuring safety has becomemore difficult for European authorities. Many of theseproducts do not enter to the internal market as largebatches, which complicates the work at customs. Inthese cases, alerts are possible but recalls are difficult. • New technologies are the final problem are recognizedby the Commission—and for a good reason. Technolo-gies such as artificial intelligence have fueled the wholesafety debate to a new level throughout the world.But according the Commission’s auxiliary report [19],the safety provisions for artificial intelligence applica-tions, robotics, and Internet-of-things devices are seenattainable mainly through transparency, accountabil-ity, and unbiasedness of algorithms, fallback mecha-nisms, and keeping a human in the loop. Althoughthese are sensible goals as such, it remains disputedhow these could be legislated with sufficient rigor. Forthese and other reasons, liability for new technologieshas been actively debated recently. Given that re-forms are already underway at this front, it suffices togenerally pinpoint the many challenges: unclear def-initions, different legal notions of liability, potential(compulsory) insurance schemes, impacts upon econ-omy and innovation, supply chains, and so on [7, 64].Another related point is the ethical use of artificialintelligence promoted by the EU. Here, it remainsunclear whether ethics can—or should—be conflatedwith safety, which should be arguably always guaran-teed regardless of what is seen as right or wrong.Three additional points deserve a brief discussion. Thefirst point is practical: the RAPEX system could bene-fit from more meta-data. For instance, the textual de-scriptions often lack sufficient information to deduce aboutwhich legislations or standards were specifically violated.Although some national authorities have occasionally usedgeneral phrasings such as “machine directive”, most of theentries lack even such elementary cues. Some criticism hasalso been expressed previously regarding missing informa-tion about specific dangerous compounds in some prod-ucts, and whether these ended to RAPEX due to produc-ers’ self-disclosures or due to testing by national author-ities [44]. Further small practical improvements are notdifficult to imagine; there are no longer working hyper-links pointing particularly to online platforms, and so on.The second point is about the accreditation and its re-lation to other ongoing policy changes. Regulation (EU)2019/881 established cyber security certification for infor-mation technology products. For the purposes of this pa-per, the certification scheme is interesting because it dif- fers from the New Approach tradition: it is not basedon harmonized standards enacted by the three Europeanstandardization organizations, and, in some cases, certi-fication via it may overlap with these standards and theproduct-specific legislations [38]. From a policy perspec-tive, further incoherence may follow in the long-run. Thisassertion leads to consider the third and final point.Last, it is sensible to contemplate whether the wholeconcept of safety should be defined better to reflect newrealities. Many of the established demarcations seem out-dated, both in academia and in practice. For instance, thedistinction between products and services has acknowl-edgedly become blurry for safety considerations [17, 19].Particularly information technology products have long re-lied on different services, nowadays often fueled by data.As the issues with global supply chains demonstrate, it isalso increasingly difficult to make clear-cut distinctions be-tween producers, suppliers, and distributors. Then, thereis the safety concept itself. The relation between secu-rity and safety has always been ambiguous. Privacy asa safety concern has also long been debated [26]. Butthere is more. Even information itself can today be seenas a safety issue—if not a hazard. For instance, recentresults indicate a substantial amount of vaccine misinfor-mation (some of which may also be intentional disinforma-tion) associated with potentially dangerous products soldon electronic commerce platforms [36]. These and othersimilar results restate the general platform problem. Butthey further call to theorize and reconceptualize safety.
References [1] Austgulen, M. H. (2020). Understanding National Preferences inEU Consumer Policy: A Regime Approach.
Journal of ConsumerPolicy , 43:767–798.[2] Ben¨ohr, I. (2020). The United Nations Guidelines for ConsumerProtection: Legal Implications and New Frontiers.
Journal ofConsumer Policy , 43:105–124.[3] Blind, K. and Gauch, S. (2008). Trends in ICT Standards: TheRelationship Between European Standardization Bodies and Stan-dard Consortia.
Telecommunications Policy , 32:503–513.[4] Boell, S. K. and Cecez-Kecmanovic, D. (2015). On Being ‘Sys-tematic’ in Literature Reviews in IS.
Journal of Information Tech-nology , 30(2):161–173.[5] Boustras, G. and Waring, A. (20). Towards a Reconceptualiza-tion of Safety and Security, Their Interactions, and Policy Require-ments in a 21st Century Context.
Safety Science , 132:104942.[6] Bradgate, R. and Twigg-Flesner, C. (2002). Expanding theBoundaries of Liability for Quality Defects.
Journal of ConsumerPolicy , 25:345–377.[7] Bugra, A. (2020). Room for Compulsory Product Liability In-surance in the European Union for Smart Robots? Reflectionson the Compelling Challenges. In Marano, P. and Noussia, K.,editors,
InsurTech: A Legal and Regulatory View , pages 167–197.Springer, Cham.[8] Bujok, A. B., MacMahon, S. T., Granta, P., Whelan, D., Rickard,W. J., and McCaffery, F. (2017). Approach to the Developmentof a Unified Framework for Safety Critical Software Development.
Computer Standards & Interfaces , 54(3):152–161.[9] Campbell-Platt, G. (1994). Food Control for Consumers, Regu-lators and Producers.
Food Control , 5(2):99–102.[10] Cavaliere, A. (2004). Product Liability in the European Union: ompensation and Deterrence Issues. European Journal of Lawand Economics , 18:299–318.[11] Clancy, C., Ferraro, J., Martin, R., Pennington, A.,Sledjeski, C., and Wiener, C. (2021). Deliver Uncompro-mised: Securing Critical Software Supply Chains: Proposalto Establish an End-to-End Framework for Software Sup-ply Chain Integrity. MITRE. Available online in January: .[12] Cordero, C. A. and Sanz, J. L. M. (2009). Measurement ofMachinery Safety Level: European Framework for Product Con-trol: Particular Case: Spanish Framework for Market Surveil-lance.
Safety Science , 47(10):1285–1296.[13] Dewitt, R. (1999). Standardisation and Legislation in Europe.
Polymer Degradation and Stability , 64(3):535–544.[14] Duarte-Davidson, R., Orford, R., Wykea, S., Griffiths, M.,Amlˆot, R., and Chilcot, R. (2014). Recent Advances to AddressEuropean Union Health Security from Cross Border ChemicalHealth Threats.
Environment International , 72:3–14.[15] Dye, T. R. (1976).
Policy Analysis: What Governments Do,Why They Do It, and What Difference It Makes . University ofAlabama Press, Tuscaloosa.[16] eBay (2020). Feedback on the Inception Impact Assess-ment on the Review of the General Product Safety Directive.Available online in January 2021: https://ec.europa.eu/info/law/better-regulation/have-your-say/initiatives/12466-Review-of-the-general-product-safety-directive/F547576 .[17] European Commission (2018). Evaluation of Council Directive85/374/EEC on the Approximation of Laws, Regulations and Ad-ministrative Provisions of the Member States Concerning Liabilityfor Defective Products. Written by EY with the support of Tech-nopolis Group and VVA. Available online in January 2021: https://op.europa.eu/en/publication-detail/-/publication/d4e3e1f5-526c-11e8-be1d-01aa75ed71a1/language-en .[18] European Commission (2020a). Combined Evalua-tion Roadmap/Inception Impact Assessment: Revisionof Directive 2001/95/EC of the European Parliamentand of the Council on General Product Safety. Avail-able online in January 2021: https://ec.europa.eu/info/law/better-regulation/have-your-say/initiatives/12466-Review-of-the-general-product-safety-directive .[19] European Commission (2020b). Report From the Commis-sion to the European Parliament, the Council and the Euro-pean Economic and Scial Committee: Report on the Safety andLiability Implications of Artificial Intelligence, the Internet ofThings and Robotics. COM/2020/64 final. Available online inJanuary 2021: https://eur-lex.europa.eu/legal-content/en/TXT/?qid=1593079180383&uri=CELEX%3A52020DC0064 .[20] European Committee for Standardization (CEN) and Eu-ropean Committee for Electrotechnical Standardization(CENELEC) (2020). Position Paper: European Commi-sion’s Consultation on the General Product Safety Directive2001/95/EC. Available online in January 2010: .[21] European Parliament (2019). Addressing Product Safetyin the Single Market: European Parliament Resolution of25 November 2020 on Addressing Product Safety in theSingle Market (2019/2190(INI)). Available online in Jan-uary 2021: .[22] European Telecommunication Standards Institute (ETSI)(2020). Calling the Shots: Standardization for EUCompetitiveness in a Digital Era. Available online inJanuary 2021: .[23] European Union (2021). Safety Gate: Rapid Alert System forDangerous Non-Food Products. Data retrieved in 19 January 2021from: https://ec.europa.eu/consumers/consumers_safety/ safety_products/rapex/alerts/?event=main.search&lng=en .[24] Eurostat (2020). Key Figures on Europe – Statistics Il-lustrated – 2020 Edition. Available online in January 2021: https://ec.europa.eu/eurostat/documents/3217494/11432756/KS-EI-20-001-EN-N.pdf .[25] Faure, M. G. (2003). Product Liability and ProductSafety in Europe: Harmonization or Differentitation.
Kyklos ,53(4):467–508.[26] Grimmelmann, J. (2010). Privacy as Product Safety.
WidenerLaw Journal , 19(3):793–828.[27] Hajer, M. and Wagenaar, H. (2003). Introduction. In
Delibera-tive Policy Analysis: Understanding Governance in the NetworkSociety , pages 1–32. Cambridge University Press, Cambridge.[28] Herzmann, K. (2015). Fundamental Rights Protection AgainstMarket-Related Information Activity: The Case of MultilevelProduct Warnings.
China-EU Law Journal , 4:33–64.[29] Hodges, C. (2004). Do We Need a European Medical DevicesAgency?
Medical Law Review , 12(3):268–284.[30] Howells, G. (2014). Product Safety—A Model for EU Legisla-tion and Reform. In Purnhagen, K. and Rott, P., editors,
Vari-eties of European Economic Law and Regulation , pages 525–538.Springer, Cham.[31] Howells, G. (2020). Protecting Consumer Protection Cvalues inthe Fourth Industrial Revolution.
Journal of Consumer Policy ,43:145–175.[32] Howells, G. G. (2000). The Relationship between Product Lia-bility and Product Safety – Understanding a Necessary Element inEuropean Product Liability through a Comparison with the U.S.Position.
Washburn Law Journal , 39(3):305–346.[33] Jin, G. Z. and Wagman, L. (2020). Big Data at the Crossroadsof Antitrust and Consumer Protection.
Information Economicsand Policy , (Published online in April):1–10.[34] Joerges, C. (2010). Product Safety, Product Safety Policy andProduct Safety Law.
Hanse Law Review , 6(2):115–134.[35] Jore, S. H. (2019). The Conceptual and Scientific Demarcationof Security in Contrast to Safety.
European Journal of SecurityResearch , 4:157–174.[36] Juneja, P. and Mitra, T. (2021). Auditing E-CommercePlatforms for Algorithmically Curated Vaccine Misinformation.Archived manuscript. Available online in January: https://arxiv.org/abs/2101.08419 .[37] Klaschka, U. (2017). Trust, But Verify! Personal Care Prod-ucts in the Rapid Alert System Database RAPEX.
SustainableChemistry and Pharmacy , 5:30–41.[38] Kohler, C. (2020). The EU Cybersecurity Act and EuropeanStandards: An Introduction to the Role of European Standard-ization.
International Cybersecurity Law Review , 1:7–12.[39] Li, Y. and Wagenaar, H. (2019). Revisiting Deliberative PolicyAnalysis.
Policy Studies , 40(5):427–436.[40] Mak, S. L. and Ke, D. M. (2017). A Review on European ToySafety Directives. In
Proceedings of the IEEE International Sym-posium on Product Safety and Compliance Engineering – Taiwan(ISPCE-TW) , pages 1–2, Taipei. IEEE.[41] Marucheck, A., Greis, N., Mena, C., and Cai, L. (2011). Prod-uct Safety and Security in the Global Supply Chain: Issues, Chal-lenges and Research Opportunities.
Journal of Operations Man-agement , 29:707–720.[42] McGregor, S. (2021). Preventing Repeated Real World AI Fail-ures by Cataloging Incidents: The AI Incident Database. Archivedmanuscript. Available online in January: https://arxiv.org/abs/2011.08512 .[43] Meltsner, A. J. (1986). The Seven Deadly Sins of Policy Anal-ysis.
Knowledge: Creation, Difussion, Utilization , 7(4):367–381.[44] Michalek, I. M., Liu, B., Benn, E. K., and dos Santos, F. L. C.(2019). Skin Lightening Products’ Violations in Europe: An Anal-ysis of the Rapid Alert System for Dangerous Non-Food Products2005–2018.
Regulatory Toxicology and Pharmacology , 106:50–54.[45] Moore, R. (2013). Standardisation: A Tool for Addressing Mar-ket Failure Within the Software Industry.
Computer Law & Se-curity Review , 29(4):413–429.[46] M´ugica, S. C. (1990). Protection of the Weak Consumer Under roduct Liability Rules. Journal of Consumer Policy , 13:299–309.[47] Oxford, R., Crabbe, H., Hague, C., Schaper, A., and Duarte-Davidson, R. (2014). Eu Alerting and Reporting Systems for Po-tential Chemical Public Health Threats and Hazards.
Environ-ment International , 72:15–25.[48] Pauwels, M. and Rogiers, V. (2010). Human Health SafetyEvaluation of Cosmetics in the EU: A Legally Imposed Challenge.
Toxicology and Applied Pharmacology , 243:260–274.[49] Peltonen, E. (2010). Could EU-Level Regulatory Private Inter-est Organisations Promote or Discourage Harmonisatio of Con-sumer Law in EU? In Tala, J. and Pakarinen, A., editors,
Bet-ter Regulation – A Critical Assessment: Proceedings from theInternational Conference on Legislative Studies , pages 141–115,Helsinki. Hakapaino Oy.[50] Playle, M. (2011). CE Marking – The Essential Requirements.In Dale, C. and Anderson, T., editors,
Advances in SystemsSafety: Proceedings of the Nineteenth Safety-Critical SystemsSymposium , pages 251–271, Southampton. Springer.[51] Purnhagen, K. (2003).
The Politics of Systemization in EUProduct Safety Regulation: Market, States, Collectivity, and In-tegration . Springer, Dordrecht.[52] Raposo, V. L. and Morbey, W. W. (2015). Product Liability inEurope and China.
China Legal Science , 3(4):60–83.[53] Rausand, M. and Utne, I. B. (2000). Product Safety – Prin-ciples and Practices in a Life Cycle Perspective.
Safety Science ,47:939–949.[54] Ruohonen, J. and Hjerppe, K. (2020). The GDPR Enforce-ment Fines at Glance. Archived manuscript, available online inDecember: https://arxiv.org/abs/2011.00946 .[55] Ruohonen, J., Hyrynsalmi, S., and Lepp¨anen, V. (2016). AnOutlook on the Institutional Evolution of the European UnionCyber Security Apparatus.
Government Information Quarterly ,33(4):746–756.[56] Ryan, D. J. (2003). Two Views on Security Software Liabil- ity: Let the Legal System Decide.
IEEE Security & Privacy ,99(1):70–72.[57] S´anchez, J. J. A., Bordin, G., and Rodr´ıguez, A. R. (1999).Sharing Knowledge in Consumer Protection: A Database on Ex-perts in Evaluation and Testing of Consumer Products and TheirRisks of Use.
International Journal for Consumer and ProductSafety , 6(4):193–204.[58] Sciascia, A. (2006). Safe or Sorry: How the Precaution-ary Principle Is Changing Europe’s Consumer Safety RegulationRegime and How the United States’ Consumer Product SafetyCommission Must Take Notice.
Administrative Law Review ,58(3):689–708.[59] Surdam, D. G. (2020).
Business Ethics from the 19th Centuryto Today: An Economist’s View . Palgrave Macmillan, Cham.[60] The European Consumer Organisation (BEUC) (2020). BEUCand ANEC Views for a Modern Regulatory Framework onProduct Safety: Achieving a Higher Level of Consumer SafetyThrough a Revision of the General Product Safety Directive.Available online in January 2021: https://ec.europa.eu/info/law/better-regulation/have-your-say/initiatives/12466-Review-of-the-general-product-safety-directive/F545598 .[61] Ullrich, C. (2019). New Approach Meets New Econony: En-forcing EU Product Safety in e-commerce.
Maastrict Journal ofEuropean and Comparative Law , 26(4):558–584.[62] van Aken, D. (2002). Assessing Risks – Challenges Posed bythe European General Product Safety Directive.
Injury Controland Safety Promotion , 9(3):151–156.[63] Weitzenboeck, E. M. (2015). Looking Back to See Ahead: TheChanging Face of Users in European e-Commerce Law.
ArtificialIntelligence and Law , 23:201–215.[64] Zech, H. (2021). Liability for AI: Public Policy Considerations.
ERA Forum , (Published online in January)., (Published online in January).