Amsterdam District Court rules on automated decision-making Uber
Uber keeps the legal community busy. In the UK, the Supreme Court recently confirmed that drivers working for Uber are not self-employed contractors, but workers and are therefore entitled to a minimum wage and holiday allowance. Similar proceedings are ongoing in France. Recently, the Amsterdam District Court ruled in two cases related to these foreign proceedings. The rulings are especially important because they address whether or not Uber takes part in an unlawful form of automated decision-making, a subject that is undeniably increasingly important, but on which little case law exists.
According to the UK Supreme Court, an employment relationship exists between the British drivers and Uber London. However, the owner of the Uber app is not the UK entity, but the Dutch parent company Uber B.V. In order to be able to appreciate the implications of the aforementioned ruling and also in preparation for other proceedings, some British drivers (case 1) therefore exercised their right of access on the basis of the GDPR against Uber B.V. They wanted to know which of their personal data Uber processes, what they are used for and to what extent profiling and automated decision-making take place.
A number of other drivers (case 2) believed that Uber had made a fully automated decision to deactivate their accounts due to suspicions of fraud, which allegedly violated the prohibition of automated individual decision-making. More specifically, these drivers were accused of falsely collecting cancellation fees from Uber by posing as both passenger and driver, or manipulating the Uber Driver app so that more expensive rides could be identified before the driver accepted the ride (which is not allowed to avoid cherry picking of rides).
The GDPR has some specific rules about automated decision-making. First of all, an organisation that wants to use automated decision-making must inform the persons concerned about this in advance. Part of this obligation is to provide meaningful information about the “logic involved” of the application. This does not require that the operation of the algorithm is explained in (technical) detail or that the algorithm is made public. However, it must be made clear in an understandable manner how decision-making works and on the basis of which criteria a decision is reached.
Furthermore, the data subject has the right not to be subject to a decision based solely on automated processing that has legal consequences for him or her or which otherwise significantly affects him / her. Although this article has been formulated as a “right”, according to the European privacy regulators, there is actually a general ban on fully automated individual decision-making, with a few exceptions. It is therefore prohibited to terminate an employment contract solely on the basis of an automated decision. Targeted advertising based on profiles is generally permitted, as this will normally not affect the data subject to a significant extent.
Uber uses a batched matching system. This system groups the closest drivers and passengers in a batch (a group) and determines the optimal match (link) between a driver and a passenger within that group. According to Uber, it uses location, direction of travel, traffic volume, geographic factors, estimated time of arrival at the passenger’s pick-up point, and personal preferences specified by drivers. The system no longer matches a passenger with a driver if this passenger has rated the driver with one of the five available stars in the past. The driver is then matched with another passenger in the batch. According to Uber, the automated allocation of available rides has no legal consequences and the data subject is not significantly affected, so that no automated decision-making takes place. The court follows Uber in this. Although it is obvious that the system will have a certain influence on the performance of the agreement between Uber and the driver, the court is of the opinion that this influence is not so great as to have a significant effect. The court dismisses the drivers’ claim.
Uber disputes that it deactivated the drivers’ accounts based solely on automated decision-making. According to Uber, this was preceded by a thorough investigation conducted by a specialised team of Uber employees. Uber does use software to detect potential fraudulent activities. In response to a fraud signal, this team investigates the facts and circumstances based on internal protocols and its own knowledge and experience in order to confirm or rule out the existence of fraud. Deactivation of an account requires a unanimous decision of two employees of this team. On the basis of this explanation, the court rules that Uber’s decisions were made after “meaningful human intervention” and that there was therefore no unlawful form of automated decision-making. This claim of the drivers is also rejected.
Would you like to know more about the legal aspects of online platforms or automated decision-making? Feel free to contact our team.
Jeroen van Helden, attorney at law IT, IP & Privacy