In the Netherlands, a recent cartel damages ruling by a court of appeal shows that a failure to manage large volumes of information can be a real issue for legal teams. Companies facing civil, regulatory or criminal proceedings have to deal with an ever-increasing amount of data. Not only the content of data, but the initial step of unlocking and categorising the data is key. Without properly executing those preliminary steps, the content is inaccessible and becomes worthless in the proceedings. Lawyers have to be well equipped to meet this new challenge, as it requires a combination of human expertise and machine learning. With predictive coding, lawyers train the algorithm to search large volumes of data. That is how artificial intelligence helps unlock content. When lawyers don't do this, they will lose their cases if the information they need stays hidden in the data. Unlocking it has become key.

In a recent follow-on cartel damages case against several lift manufacturers, customers bundled their claims for damages in a claim vehicle, East West Debt B.V. (EWD). As a claim vehicle, EWD had a duty to substantiate its claim. The Arnhem-Leeuwarden Court of Appeal gave EWD the opportunity to submit, per party, a file providing more detailed information on each claim. Instead, EWD submitted a USB stick with digital files containing over 15,000 pages of information. The court of appeal did not look at the content. Instead, it held that EWD had not sufficiently substantiated its claim, as the documents were unorganised and EWD had not provided explanations for the information. According to the court, EWD had not sufficiently substantiated its claims. As a result, EWD lost the case. As many who have worked on similar types of cases can attest, 15,000 pages is not bad, and often covers a small amount of information. For example, the file in the Dutch fruit juices and concentrate cartel contained approximately 49,000 documents. When the European Commission prohibited Siemens' takeover of Alstom, it analysed over 800,000 documents, including internal documents. And in Bayer's acquisition of Monsanto, the Commission even reviewed 2.7 million internal documents. In comparison, 15,000 pages does not seem excessive. Large amounts of data can delay the outcome of a case. The data must not only be properly unlocked, it also takes opposing parties more time to respond to statements filed by the other party based on those data, as they need to review the data first. In an example of how this can be detrimental to opposing parties, in a 2018 case, the UK competition authority required companies in a merger control procedure to respond to data totalling almost 9GB of data (and 163 separate sets of codes with econometric analysis of those data), within a short (and unreasonable) period of time. In that case, the Competition Appeals Tribunal held that the authority had infringed the companies' rights to a proper defence. And it's not only lawyers and authorities who will have to learn how to take on this new digital task: judges will, too. In the "Amsterdam Passage process", the criminal file consisted of 600 binders full of documents. This is probably one of the last cases on paper, since judges in criminal cases work more and more digitally. Of course, the looming question is how to properly manage this large volume of data, since simply printing and reading the documents is becoming impossible. On the other hand, computers alone cannot conduct a data review . Data review is a combination of human expertise and machines. A team of experts – for example, lawyers or forensic accountants – must review the data digitally by using keywords which they are familiar with professionally. Via predictive coding, the experts train the algorithm, which is then able to restart the search of the data set repeatedly as needed. That is how artificial intelligence helps unlock large volumes of data. While doing so, safeguards have to be built into the process. This includes not only safeguards with respect to legally privileged documents but also safeguards to protect the privacy of the persons mentioned in the data, especially when data is transferred outside the EU in the course of proceedings. For example, in an e-discovery case before a US court.