Making Algorithm Registers Work for Meaningful Transparency

More and more algorithms are used in critical fields of the public sector. Yet, even as some of these systems have been shown to have dire consequences on citizens, their use remains opaque. The European Union’s recent AI Act introduces an EU-wide database, managed by the European Commission, in which high risk AI systems will have to be registered publicly. This includes algorithmic systems used in critical areas of the public sector, such as education, access to public services, or justice, and in the private sector, such as banking and insurance. The database is not devoid of limitations. However, it is an opportunity for governments to tackle the topic of algorithm registers at a national level, and use them as a tool to ensure transparency, accountability and the safeguarding of fundamental rights in the way algorithmic systems are designed, developed, implemented and assessed. Against this backdrop, this report aims to inform governments wishing to develop an algorithm register in their national context, and organizations advocating for their development. It attempts to answer the question: how to design, implement, and evaluate a national algorithm register that serves transparency, accountability, and the safeguarding of fundamental rights?