Barcelona approves an internal protocol to introduce AI into municipal services with ethical guarantees

The Barcelona City Council approved the “Definition of work methodologies and protocols for implementing algorithmic systems”, which creates an internal protocol for the ethical implementation of algorithmic systems used by the municipality. The protocol, based on the proposed European AI regulation, defines step-by-step mechanisms for each stage of the process, from the public tendering and implementation phase to the dismantling of the AI system. The procedure adapts to the different stages of the AI lifecycle and its level of riskiness (assessed from ‘limited and minimal risk’ to ‘unacceptable risk’ based on the EU Commission’s risk classification). According to this classification, algorithmic systems with unacceptable risk are automatically rejected, while high-risk systems must undergo an algorithmic impact study produced by an external consultant. Furthermore, it also establishes governance and supervision bodies that will ensure that the impact of AI is in line with the ethical principles endorsed by Barcelona’s AI strategy. This is a pioneering document in regard to the regulation and governance of algorithmic systems at the local level, as it combines public procurement, the lifecycle of an algorithmic system, the City Council’s data protection mechanisms and the ethical standards defined by various administrations and organisations of the city.