AI applications are entering a new phase within the judiciary. They are already being trialled in the anonymisation of court decisions, the processing of large volumes of cases, research, the structuring of legal submissions and the support of procedural workflows. However, successful pilot projects do not yet translate into reliable, scalable and legally sound routine operations.

This brings practical implementation to the fore: the crucial question is which technical, organisational and legal requirements must be met to ensure that AI applications can be used on a permanent basis. At the same time, a clear distinction must be made between where systems merely assist, where they automate processes, and where they pre-structure or influence judicial decisions. Particular challenges arise where applications are classified as high-risk systems.

The forum asks:

• Which AI applications are already working in judicial practice – and what are their limitations?
• What is needed to roll out pilot projects safely on a wider scale?
• How can oversight, transparency and human accountability be organised in day-to-day work?
• And when does the use of AI in the judiciary become a high-risk case – and what are the consequences for development, procurement and operation?