The rise of machine learning gives rise to knotty legal problems, whether these relate to how the law applies to robots and artificial intelligence*, or to algorithms which may be designed by humans, but then continue to run and 'evolve' automatically. For competition lawyers, a current topic of debate is whether machines are capable of collusion in a way which engages (or, potentially, evades) antitrust liability (see, for example, Ariel Ezrachi and Maurice Stucke's 2015 article "Artificial Intelligence and Collusion: When computers inhibit competition".
It is likely to be some time before there is anything like a definitive answer to such questions. However, a recent Court of Justice ruling has something of this flavour, and is an important judgment for companies doing business online.
Eturas, the case in question, concerned a platform which was used by travel agents wishing to have an online presence without the need to develop their own website. The CJEU has now endorsed the view of the Advocate General (on which we reported last year) that an anti-competitive agreement is capable of arising between members of a platform an its administrator when platform terms and conditions are updated. This is so even if those companies had no direct knowledge of the acts of the other platform members, nor any direct contact with them - rather like the 'hub and spoke' type agreements identified between suppliers and separate distributors in a number of old OFT cases.
The update in question in this case concerned the introduction of a cap on discounts which could be offered by platform members. Because an agreement which limits independent companies' ability to offer discounts on their separate (but competing) commercial offerings clearly infringes Article 101(1), the questions referred to the Court of Justice focussed on whether an agreement or concerted practice could be said to exist in circumstances where the participating companies had not given their express approval for the change in the T&Cs. The Court treated this as a question of the standard of proof: if the travel agents were aware of the administrator's message they may be presumed to have participated in an unlawful concerted practice unless they have publicly distanced themselves from the change, or have taken other steps sufficient to rebut the presumption of agreement.
Much of this will be a matter for national law, subject to – on the one hand – the presumption of innocence and, on the other, the principle of effectiveness which must be observed by national legal systems when applying Treaty provisions. However, the Court made clear that national courts should not require that "excessive or unrealistic" steps be taken by companies to distance themselves from such an infringement. Equally important is the response to the question of whether companies "ought to have been aware" of the infringement. The judgment (taking a slightly narrower approach than that suggested by the Advocate General) requires actual awareness for an infringement to be established, albeit that it may be possible, as a matter of evidence, to infer this from particular indicia which are suggestive of such awareness.
All in all, the Court has taken a relatively cautious approach. While the floodgates to liability have not been opened, there will no doubt be more such cases in future. In this case, the change to the T&Cs was sent by direct email to the platform members, but in principle an automated change could engage the same issues. To avoid being one of those future cases, careful consideration should be given to the scope for excluding such liability in contractual terms agreed with platform providers and, in particular, how such relationships are managed at an operational level. It will not necessarily be the case that personnel who normally deal with such matters will have the relevant awareness of the competition rules, so appropriate training and/or supervision may need to be provided.