Nine Riis is a PhD Fellow at the Centre for Private Governance at the Faculty of Law, University of Copenhagen.
1. Introduction
We have all heard the repeated exclamations of data being the new oil and most people with an eye on the lawmakers in Brussels know that effective use of data is a top priority for the Commission. To aid the EU data market, one of the initiatives in the European strategy for data is to ensure better access to privately held data through B2B contracting. A main obstacle in this regard is that companies (especially SMEs) experience strong market players abusing their bargaining power by supplying data on “take-it-or-leave-it-terms”. Such terms deter counterparties from using the provided data efficiently and from entering into data contracts at all. In response, the Commission has proposed an unfairness test in art. 13 of the proposed EU Data Act. There is, however, a genuine concern that art. 13’s vague and ambiguous wording leads to legal uncertainty disincentivising data trade contrary to the provision’s objective.
This blog post briefly explains what a data contract is, offers an overview of art. 13 of the proposed EU Data Act and analyses the concerns it gives rise to as well as potential amendments.
2. What are data contracts?
Data contracts can broadly be defined as any type of contract wholly or partially regulating the access to and/or use of data. Data contracts are a relatively new phenomenon meaning that the development of standardised clauses and a uniform understanding of market practice is still in its infancy. Thus, there is not yet any standardised way of drafting data contracts. Efforts devoted to systematisation so far indicate that one of the most common ways to approach data contracts is to adopt a licensing model (e.g. ALI-ELI Principles for a Data Economy and the Danish Business Authority’s template). A licensing model consists of the data supplier granting the data recipient a right to use the provided data under specific terms. A licensing model can cover many different types of data exchanges between companies. One of the simplest setups is when a company wishes to develop an algorithm. In such case, the company needs training data for development. If the company does not possess such training data itself, it can be procured from another company via a data contract specifying the terms on which the data can be used and the remuneration payable for it. For example, a clause could be: “Company A is granted a time-limited, non-exclusive right to use Data Set X to train Algorithm A. Company A can solely use Data Set X for the purposes set out in this Agreement and upon payment of 1,000 EUR per month”.
A more complex situation arises when a company generates data as a by-product of another activity, for example, through use of a software programme accessed through a software license contract. Such data is potentially valuable both to the company generating it, and also to the company supplying the software programme as the latter can use it for further development. Accordingly, the software license contract is likely to regulate issues such as who can use the data and how (akin to the provision worded above). By partially regulating data, the software license contract also falls within the above-stated definition of a data contract. Many other scenarios for exchanges of data exist (including with several parties) and the aim herein is not to provide an exhaustive list, but to illustrate that exchanges of data (and thus the design of data contracts) can take many different forms.
3. Art. 13 of the proposed EU Data Act – an unfairness test for contractual terms
B2B parties have had almost unlimited contractual freedom when concluding data contracts and any restrictions have been confined to sector-specific regulation and restrictions in national contract law. It is relevant to note that several Member States employ a general unfairness test in B2B relations (see e.g. the list on p. 76). By way of illustration, § 36 of the Danish Contracts Act stipulates that a contract can be wholly or partially set aside if it is unfair or contrary to honest conduct to enforce it.
Art. 13 of the proposed EU Data Act modifies the outset of contractual freedom by setting out an unfairness test for certain terms in data contracts. The scope of art. 13 is broad as it applies across sectors. Pursuant to art. 13, a contractual term unilaterally imposed on an SME regarding (i) access to and use of data or (ii) the liability and remedies for the breach or the termination of data-related obligations shall not be binding on the SME if it is unfair. Art. 13 includes both a general unfairness test (art. 13(2)) as well as a list of contractual terms always considered unfair (art. 13(3)) and a list of terms presumed unfair (art. 13(4)).
The general unfairness test states that a contractual term is unfair if “[…] it is of such a nature that its use grossly deviates from good commercial practice in data access and use, contrary to good faith and fair dealing”. Recital 54 of the proposed EU Data Act specifies that the test only applies to “[…] excessive contractual terms, where a stronger bargaining position is abused” and not when a term merely favours one party over the other. Accordingly, the provision targets take-it-or-leave-it situations where a disproportionate burden is placed on the weaker party (p. 166 of the impact assessment), as such provisions disincentivise data trade. In light of SMEs’ experienced difficulties in obtaining access to data (p. 18 of the impact assessment), the rationale of the provision is sound. However, it is problematic that the benchmark for the unfairness test relies on the assumption that an established market practice for data contracts exists when data contracts are just beginning to emerge. Presumably, this partially explains why art. 34 of the proposed EU Data Act stipulates an obligation for the Commission to draft non-binding model contractual terms. While efforts to standardise data contracts are welcome, difficulties arise when drafting clauses that can function as a benchmark for unfairness across sectors. While recital 83a recognises this by stating that the model terms shall take “[…] into account the conditions in specific sectors […]”, it is unclear how it will be solved in practice. Further, it is not specified to what extent market players will be involved in the drafting of such model terms. Consequently, the vague wording of the general unfairness test leads to legal uncertainty harming data trade as companies might either decide against trading data at all or raise prices for the traded data to mitigate the effects of the unfavourable terms.
Art. 13 also delimits the concept of unfairness in the list of terms always considered unfair and presumed unfair. The terms always considered unfair are:
- exclusion or limitation of liability for intentional acts or gross negligence,
- exclusion of remedies available to the party upon which the terms has been unilaterally imposed, or
- giving the party imposing the term the exclusive right to determine whether the data supplied are in conformity with the contract or to interpret any term of the contract.
Within Danish contract law such provisions are likely invalidated pursuant to § 36 of the Danish Contracts Act (see e.g. U.2010.3113.H recognising invalidity of limitation of liability in cases of gross negligence). The logic of the rule is that companies should not have a free pass to deliver bad-quality services without liability. Yet, in contrast to art. 13(3), § 36 still requires the specific circumstances of the transaction to be taken into account (see § 36(2)), thus it is not a per se prohibition.
It is important to note that § 36 of the Danish Contracts Act does not necessarily invalidate the terms presumed unfair listed in art. 13(4). The presumption of unfairness needs to be rebutted by the company unilaterally imposing the contractual term. However, this may prove difficult due to the vague wording of the article. For instance, art. 13(4)(a) states that a contractual term is presumed unfair if its object or effect is to “inappropriately limit the remedies in case of non-performance of contractual obligations or the liability in case of breach of those obligations”. It is common practice to limit the use of remedies in B2B relations, and it is therefore difficult to determine when such limitation is “inappropriate”. As the complete exclusion of remedies is prohibited according to art. 13(3), the provision must cover cases between complete exclusion and all remedies being available, which is a difficult middle ground to identify. It can be noted that Danish courts often emphasise that B2B agreements are based on the parties’ calculations of commercial and financial risks (see e.g. U.2012.3007.H) and that an agreement cannot be invalidated pursuant to § 36 merely because it is (or turns out to be) unfavourable. A commercial party must take responsibility for its commercial decisions and cannot expect to be subsequently saved by the law. A similar argument can be made in relation to art. 13(4)(a).
Along the same lines, art. 13(4)(c) states that a contractual term is presumed unfair if its object or effect is to “prevent the party upon whom the term has been unilaterally imposed from using the data contributed or generated by that party during the period of the contract, or to limit the use of such data to the extent that that party is not entitled to use, capture, access or control such data or exploit the value of such data in a proportionate manner”. There is no benchmark for what constitutes “proportionate exploitation of the value of data”. While the development of directly competing products is likely excluded, it is unclear if the data can be used to develop a competing product within another market where the supplier of the provided data is also active. Art. 13(7) also excludes the application of the unfairness test to the adequacy of the price paid for the data. On its own, this provision is sensible as it allows for free pricing. Yet, in conjunction with art. 13(4)(c) it is problematic as price often is a vital part of contractual negotiations and correlates to the scope of rights afforded the party accessing the data. From the current wording, it is unclear if the price payable for the data is a valid consideration to include when determining proportionate exploitation. Consequently, both art. 13(4)(a) and (c) give rise to considerable legal uncertainty which risks disincentivising data trade.
4. Concluding remarks
Summarising, the rationale of art. 13 is on the one hand sound due to SMEs’ experienced difficulties in accessing data. The clauses always considered unfair clearly invalidate terms completely excluding liability and though national contract law may already cover this, harmonisation is desirable to ensure consistency and a per se prohibition. On the other hand, an overarching concern is the vague wording employed both in the general unfairness test and the terms presumed to be unfair. The ambiguity increases legal uncertainty with an ensuing risk for data trade. However, this concern can be alleviated by removing the general unfairness test and more specifically delimiting the terms presumed unfair. As an inspiration, the Food Supply Chain Directive – one of the few other EU legal instruments limiting contractual freedom in a B2B context (see also p. 72) – does not employ a general unfairness test, but in detail set out prohibited unfair practices (see art. 3), thus clearly targeting the conduct problematic within the sector. A similar solution can be created in art. 13 to ensure that only the clearly harmful practices experienced by SMEs in data markets across sectors are prohibited. Such an approach would pursue the provision’s objective of eliminating abuse of bargaining power while ensuring that companies are aware of the terms considered unfair. Thus, the risk of companies not entering into data contracts or raising prices to mitigate unfavourable terms is minimised. Hopefully, this will be a focal point in the upcoming negotiations.