Regulating Gambling-Like Video Game Loot Boxes: a Public Health Framework Comparing Industry Self-Regulation, Existing National Legal Approaches, and Other Potential Approaches

Loot boxes are presently frequently implemented in video games, particularly on mobile phone platforms: in 2019, amongst the highest-grossing video games, 59% on the Apple iPhone platform contained loot boxes in the UK, as did 36% on the PC Steam platform [23•]. Loot boxes remain an evolving issue as the prevalence rate was found to have increased to 77% for UK iPhone games when it was re-assessed in 2021 using a comparable sample [24]: this appears to be due to multiple reasons including (i) a greater number of popular games starting to implement loot boxes; (ii) difficulties with identifying well-hidden loot box implementations with complex purchasing procedures; and (iii) semantic and definitional ambiguities with what game mechanic exactly constitutes a loot box [25•]. The prevalence rate also differs across countries: 91% of the 100 highest-grossing iPhone games contained loot boxes in China in 2020 [26], suggesting that the loot box situation might be different across countries and cultures and that future research should include the perspectives of non-Western countries and players [see 27].

Although some members of the public, e.g., some parents [see 28], consider loot boxes to be unsuitable for children, game companies and self-regulatory video game age rating systems, which are financially supported by the industry and therefore arguably conflicted from acting against the industry’s commercial interests, generally deem loot boxes to be suitable for implementation in children’s games and, by implication, suitable for children to purchase (e.g., the Entertainment Software Rating Board (ESRB) in North America and Pan European Game Information (PEGI) in Europe, neither of which requires a minimum age rating for games containing loot boxes as of May 2022, although imposing this would be within their self-regulatory powers [29]; this should be contrasted with how mere depiction of tobacco, alcohol and illegal drug use would generally render the game to be deemed as suitable only for older adolescents, e.g., ‘PEGI 16’ or suitable for players aged 16 and above [30]). Indeed, 58% of the highest-grossing UK iPhone games deemed suitable for children aged 12 + contained loot boxes in 2019 [23•], meaning that children are regularly exposed to loot boxes and can readily purchase them. The UK Gambling Commission found that 23% of young people aged between 11 and 16 have paid real-world money to purchase loot boxes [31(p. 39)]; in contrast, only 7% have ever participated in traditional online gambling [31(p. 33)].

Purchasing loot boxes is conceptually similar to gambling both structurally and psychologically because the player voluntarily spends real-world money to engage in a randomised process whose results could be desirable or, more often, undesirable, given that most potential loot box rewards are often contextually worthless to the player (either because they already have a duplicate copy or because they are already in possession of some other superior or effectively equivalent in-game item) [2•, 32]. Through purchasing loot boxes, players potentially either ‘gain’ by obtaining a valuable and wanted item or ‘lose’ by obtaining a non-valuable and unwanted item [32]. This is even more evident in relation to loot boxes that provide rewards possessing real-world monetary value that the player can subsequently sell on the secondary market, because many non-valuable rewards are worth far less than the cost of purchasing the loot box, compared to the potential large ‘jackpot’ wins from valuable rewards [3], similar to the incentive structure of lottery tickets [24]. Certain particularly rare and highly sought-after loot box content is worth hundreds, and potentially over one thousand, euros on the secondary market [33]. Indeed, opening ‘rare’ rewards from loot boxes elicits physiological responses similar to participating in certain traditional gambling activities [34]. One adult player reportedly spent over US$10,000 on loot boxes in one game over a 2-year period [35], and four children spent ‘nearly £550 in 3 weeks’ of their father’s money without permission and still failed to obtain the rare item that they were hoping for [36].

Importantly, loot box expenditure has been found to be positively correlated with self-reported problem gambling severity in many cross-sectional studies across various Western countries, including the USA [37], Spain [38], Denmark [39] and Australia [40], amongst both adult and adolescent player samples [41]. Reviews and meta-analyses of these studies have been conducted elsewhere [42••]; [43‐45]. However, the causal direction (if any) of this positive correlation is not known [46], and there is debate as to how the current evidence base should be interpreted. McCaffrey has argued that, presently, there is insufficient evidence to demonstrate that loot boxes cause widespread harm and that regulatory intervention is therefore not yet justified [47, 48]. In contrast, Drummond et al. have argued that there is already sufficient evidence demonstrating the potential harms of loot boxes (particularly, similarities with traditional gambling and the involvement of real-world money) and that loot boxes should therefore immediately be regulated more stringently [2•, 3].

These diverging perspectives have similarly been present in the longer-standing debate in gambling. Collins et al. have argued that existing harm reduction methods have been successful (as evidenced by stable or flat prevalence rates of problem gambling), and that further regulation would unnecessarily reduce the (safe) enjoyment of gambling as a leisure activity by many people [49(p. 994)]. The gambling industry also echoes this interpretation [e.g., 50], perhaps unsurprisingly given its commercial interests. In contrast, other researchers have argued that the problem gambling prevalence rate fails to reflect the full extent of the potential harms of gambling and that non- ‘problem gamblers’ would also potentially suffer harms, which is why a population-based public health harm minimisation approach might be required to reduce the risk of harm amongst all gamblers [51‐56].

The conflicting interpretations, in both the loot box and the gambling contexts, arguably arise partially from the methodological weaknesses of a majority of the evidence base, which relied on retrospective self-reported data. Such data, derived either from a representative panel (such as a prevalence survey [57]) or from online convenience samples [58], might lack reliability due to the participants’ responses being intentionally dishonest (due to a desire to hide one’s participation in gambling due to perceived stigma [59]) or unintentionally inaccurate (due to memory recall issues [60]; inconsistent interpretation of questions [61]; or incorrect estimations and calculations of expenditures [62]). To illustrate, in the UK, gambling prevalence studies have, in the 12 months prior to May 2022, recorded rates of problem gambling as varied as 0.2% (by the UK Gambling Commission [63]) and 2.8% (by YouGov on behalf of GambleAware) [64(p. 2)]. This suggests that ‘harm prevalence’ data may be subject to a range of methodological issues affecting their accuracy and validity that researchers are not fully aware of and cannot account for. More recently, Muggleton et al., relying instead on more objective transaction data from a high street bank, suggested that even the higher, previously identified problem gambling prevalence rates (i.e., ~ 3.0%) have underestimated the extent of gambling-related harms and, importantly, failed to reflect the widespread associations between gambling and various harms even amongst less engaged (supposedly, non-problem) gambling participants [65]. For loot boxes, this suggests that the prevalence of loot box ‘harms’ might have been underestimated by prior studies and that research collaborations with the industry using players’ actual (and, therefore, more objective) loot box spending data could provide important insights [66]. However, in yet another striking parallel with gambling research, the video game industry has hitherto been unwilling to share their data, similarly to how the gambling industry has been unwilling to share data with independent researchers [67].

The video game industry has adopted certain pre-emptive self-regulation purportedly to enhance transparency and reduce harms ahead of potential impending legal regulation. For example, some companies have committed to making probability disclosures detailing the player’s likelihood of obtaining different randomised rewards voluntarily outside of Mainland China (where, uniquely, disclosures are required by law [26])[68]. Major app stores, such as the Google Play Store and the Apple App Store, also require probability disclosures globally [69, 70]. In addition, the two major self-regulatory age rating systems of North America and Europe, the ESRB and PEGI, have introduced an ‘in-game purchases (includes random items)’ content descriptor to label and signify the inclusion of loot boxes in a video game [71, 72]. However, this self-regulatory measure has been criticised as being insufficiently detailed to truly inform potential customers about the risks involved with loot boxes [29], and there is no evidence of these labels providing any tangible benefit.

Loot box-related industry self-regulation mirrors many attempts by the traditional gambling industry to self-regulate (seemingly in conflict with its own financial interests) arguably in order to fend off stronger (and likely more effective) interventions from regulators and policymakers [73]. For example, the gambling industry has, for a long time, directed funds towards non-restrictive interventions, such as warning messages or education programmes about the risks of gambling, which do nothing to alter the properties and availability of potentially harmful products [74]. In the UK, the industry has agreed to partial restrictions around advertising in professional sports (e.g., ‘whistle-to-whistle’ ban): however, this has not effectively reduced sports watchers’ frequent exposure to gambling marketing via logos and website addresses shown on, for example, shirts and pitch-side billboards in soccer [75]. Research has additionally critiqued the inadequacies of the main gambling warning message used by the UK industry from 2014 to 2021 (‘When the Fun Stops, Stop’) [76, 77], and also its lack of effect on influencing gambling behaviour [78]. Many gambling researchers have therefore advocated for stronger interventions, akin to the graphic health warnings or restrictions on availability adopted in tobacco contexts [79], but uptake of these approaches has been slow.

As mentioned under “Loot Boxes and Gambling Both Show a Trend Toward Pre-emptive Industry ‘Self-Regulation’” section, the video game industry self-regulates loot boxes through mandating probability disclosures. However, importantly, the motivations for the industry to self-regulate should be viewed with an appropriate degree of scepticism because established research from traditional addictive domains, such as gambling, alcohol and tobacco, have all suggested that, when their industries have purported to act in socially responsible ways, those self-regulatory efforts have in fact been self-interested and suboptimal and have therefore failed to maximally advance the public interest [101‐103]. Indeed, industry self-regulation might not have been adopted by companies for purely altruistic purposes (e.g., improve public welfare at the cost of its own commercial profits) and may instead have been adopted to placate public concern, dissuade stricter legal regulation, and maintain control over whether and how much the product is regulated [104].

This cynical view is justified in relation to the self-regulation of loot boxes. The Apple App Store requires loot box probability disclosures for all video games on the platform. However, when the 100 highest-grossing UK iPhone games were examined in 2021, only 64.0% of those games containing loot boxes actually complied with industry self-regulation and disclosed probabilities [24]. Additionally, many UK probability disclosures were found to have been implemented using methods that were difficult for players to access (e.g., requiring multiple buttons to be pressed before the disclosure is shown) [24]. Despite this unsatisfactory level of compliance (which is likely reflected also in other countries adopting industry self-regulation), the industry’s adoption of probability disclosure self-regulation has been widely promoted: however, this measure’s effectiveness (particularly in relation to children) is not even known, and research from other risk communication domains and self-reported evidence would suggest that it is unlikely to reduce loot box spending on a broad scale [93••, 105]. This perfunctory and unsatisfactory state of affairs is reminiscent of similarly suboptimal information disclosure-based industry self-regulation in gambling [78] and other addictive domains, e.g., tobacco [101].

In contrast to how probability disclosures have been required in other countries through industry self-regulation, China has imposed this measure by law [26]. A direct comparison of the loot box probability disclosure rates amongst the 100 highest-grossing iPhone games in China and in the UK has been conducted: the compliance rate with Chinese law was 95.6% in 2020 [26], which was significantly higher than the compliance rate with UK industry self-regulation at 64.0% in 2021 [24]. This demonstrates that legal regulation appears to have been more effective at ensuring compliance than industry self-regulation (cultural differences between the two countries as to companies’ willingness to comply with law and regulation notwithstanding) [24]. However, Chinese law gave discretion to companies as to how they can comply: any disclosure, however difficult for the player to access, is deemed compliant [106]. For this reason, many disclosures in China were also found to have been published by video game companies using methods that were not prominent and difficult for players to access: even though companies could have displayed the probability disclosure on the in-game page where loot boxes could be purchased, so that players can easily view them and perhaps make more informed purchasing decisions, only 5.5% of games containing loot boxes did so [26]. In one extreme example, the player had to enter the Chinese game’s settings menu and chat with the customer support bot in English in order to access the disclosure [26]. Further, the effectiveness of probability disclosures at reducing overspending and harm is unproven and doubtful, even when they are easily accessible and have been seen by the player: only a small minority of Chinese players (16.4%) self-reported spending less money after seeing loot box probability disclosures [93••, 105]. Thus, it is important not to treat the adoption of only one consumer protection measure as a ‘solution,’ given that the measure might not be complied with fully and that the measure itself might not effectively reduce harm even if it has been effectively adopted [93••].

In other countries, although a consumer protection law approach to loot box regulation (e.g., using the Unfair Commercial Practices Directive 2005 and national implementations thereof in the European Union and the UK [106, 107•, 108, 109]) and other approaches might be tenable, the focus has been to apply gambling law to regulate loot boxes in light of the apparent similarity and relationship between loot boxes and gambling, and the ease and promptness with which existing gambling law could be applied to immediately address the issue [22]. This assessment has already been completed by the national gambling regulators of many countries, inter alia, the UK [15], the Netherlands [16] (which has since been found to be incorrect [19, 20]), Belgium [13], France [110], and Denmark [14]. The national gambling regulators would attempt to fit various implementations of loot boxes within the pre-existing national gambling law framework, meaning that the legal definitions of ‘gambling’ differ from the common sense understanding of gambling. National gambling laws also differ across countries: specifically, the various legal elements that must be satisfied for a product to constitute gambling are not the same [5•, 22]. This means that different national regulators may easily arrive at different conclusions as to whether a specific type of loot box legally constitutes gambling under the national laws of any particular country. This also means that the conclusion reached by any one national regulator (one way or the other) is not necessarily reflective of, or relevant to, the decision that a different country’s regulator might arrive at [20]. The decision-making processes are separate and based solely on how that country’s gambling law was originally drafted. Finally, national gambling regulators generally cannot change the law (from how it was originally drafted by the legislature) or make new laws: they merely pronounce an interpretation; express a desire to enforce that interpretation; and potentially take enforcement actions against products that are deemed to be contravening the law (with the proviso that the regulators’ interpretation of the law might be legally wrong and therefore remains challengeable in court by video game companies, as has been successfully done by Electronic Arts against the Dutch gambling regulator’s previously published interpretation [19, 20]).

Briefly put, Belgium, due to the distinctiveness of its national gambling law, has deemed the third and fourth categories of loot boxes (as described in Table 1; both of which require payment of real-world money to engage in a process that provides randomised rewards) to legally constitute gambling [13] and therefore effectively ‘banned’ all implementations of the product from the country [22, 100]. In contrast, most other countries (e.g., the UK [15], France [110], and Denmark [14]) concluded that only loot boxes that both require payment of real-world money to purchase and provide players with rewards that can be transferred to other players in exchange for real-world money (i.e., only the fourth category of loot boxes per Table 1) legally constitute gambling. Notably, although the countries adopting this latter position agreed as to which category of loot boxes legally constitute gambling under their laws, the national regulators’ enforcement actions have differed in relation to the same games that arguably contravene the gambling laws of multiple countries. Indeed, only the Dutch gambling regulator was known to have enforced the law [16‐18], whilst the regulators of many other countries have chosen not to act despite having issued compliance advice suggesting that such loot boxes would be illegal [e.g., 15]. However, a recent Dutch judicial decision overruled the Dutch gambling regulator’s interpretation of the law [19] and instead effectively affirmed the legality of the fourth category of loot boxes in the Netherlands [20].

Accordingly, no country, besides Belgium, is actively regulating loot boxes using gambling law at present. Some companies quickly sought to comply with Belgian law by changing the design of the national version of the game, specifically removing the possibility of purchasing loot boxes with real-world money [111, 112]. Doing so allowed these games to continue to be available to Belgian players: importantly, the games were not banned, and only the loot box monetisation method was. These corporate actions also suggest that (i) it is possible to rapidly remove the loot box functionality if required to do so and (ii) these games were still deemed to be commercially sound even without the loot box revenue stream. This might be due to these games being able to generate revenue through the sale of the software or other non-randomised in-game product offerings, or because maintaining strong brand awareness amongst the player base in Belgium was deemed as being worth the loss in revenue and operating costs. However, in contrast, other companies instead removed their games from the Belgian market entirely [113], rather than to only remove the loot box feature, likely because it was no longer commercially viable to operate those games. This shows that some genres of video games (so-called gacha games [114]) whose monetisation models rely heavily on loot boxes were likely more severely affected by the Belgian ban. The effectiveness of Belgium’s blanket ‘ban’ of loot boxes remains to be assessed, although this measure appears to have not perfectly achieved the elimination of paid loot boxes from that market [99].

Many regulatory measures that may be taken in relation to loot boxes are presented on Table 2. Notably, many are phrased as an intervention that can be imposed on players. This framing seemingly places the burden on individuals to change their loot box purchasing behaviour, which perhaps is inequitable because it is the video game companies that are providing a potentially harmful product, so it is their behaviour that policy should aim to influence. A public health approach to gambling regulation has recognised the importance of moving past the ‘blame-the-victim’ framing that is inappropriately preoccupied with the gamblers’ individual responsibility [54]. Therefore, the loot box regulatory measures could also be rephrased as interventions against video game companies, e.g., prohibiting the ‘sale’ of loot boxes by companies, instead of prohibiting the ‘purchase’ of loot boxes by players, or restricting the amount of money that ‘companies are allowed to receive’ from each player, rather than limiting the amount of money ‘players are allowed to spend.’ This would more accurately reflect that the policies are aimed at targeting the product availability that companies provide, rather than restricting players’ ability to purchase, although practically the two might be identical. The responsibility should rest with the companies to do less harm, rather than for players to protect themselves, and the framing of any regulation should more accurately reflect with whom that onus lies.

In addition, there are other potential ways of minimising loot box harms that emulate examples from public health issues other than gambling. For example, promoting the use of e-cigarettes in smoking is a way of reformulating the delivery of nicotine in a way that is fundamentally less harmful than combustible tobacco [89, 90]. Similarly, less harmful and fairer loot boxes could be implemented by increasing the likelihood of obtaining the rarest rewards; limiting how many different loot boxes may be offered within a single game and how many different potential rewards may be obtainable from a single type of loot box; and not providing players with useless (or significantly devalued) duplicate rewards [93••]. Importantly, such design changes are plausible because of one fundamental difference between loot boxes and traditional gambling: gambling providers profit only when gamblers lose money and so harm to gambling participants is inherent to the profitability of that industry; however, loot box providers profit from each loot box sale regardless of whether the video game player has ‘won’ a valuable reward or not [93••] and so profitability is not dependent on the player ‘losing’ money and harm is not inherent to the loot box monetisation model. Video game companies’ commercial interests might be negatively affected by the adoption of the aforementioned measures, as players would now only need to buy fewer loot boxes before becoming satisfied with their rewards. However, the business model would still in theory be potentially profitable (and previously non-spending players may now be more willing to spend small amounts of money on the game as the chance of obtaining a valuable item would be higher, thereby unlocking a new source of revenue for video game companies). Similar design changes would be impossible for traditional gambling as the industry would be rendered unprofitable (the house edge would be lost) [93••]. Such design-based approaches are, in the authors’ opinion, the best regulatory proposal at present because it balances the interests of all stakeholders: the potential harms of loot boxes would be effectively curtailed, but players and companies would still gain from the economic benefits of the loot box monetisation model.

The video game industry and individual companies should be encouraged to self-regulate and adopt so-called ethical game design measures (even potentially through granting tax incentives for making more ‘ethical’ games or placing tax disincentives on loot box purchases) [119]. However, any one particular measure on its own should not be deemed as sufficient consumer protection. The effectiveness of self-regulation should be continually monitored, and some legal intervention (e.g., banning certain problematic aspects of loot boxes) might be appropriate, if voluntary measures are shown to be merely performative and ineffective. Similarly, any legal regulatory measures that have been adopted (e.g., Belgium’s ban on paid loot boxes and China’s probability disclosure requirements) should also not be assumed to be an ultimate and effective ‘solution’ that other jurisdictions should immediately emulate without question: the differing cultural contexts should be considered. Which approach a certain jurisdiction decides to take is a policy decision for the people of that jurisdiction to make (national loot box research would allow for evidence-informed regulation) [80]. Consumer protection regulation can always be improved upon and must continue to adapt, as loot boxes continue to be designed and implemented in newer ways by video game companies.