The AI Act in a law enforcement context: The case of automatic speech recognition for transcribing investigative interviews

Abstract

Law enforcement agencies manually transcribe thousands of investigative interviews per year in relation to different crimes. In order to automate and improve efficiency in the transcription of such interviews, applied research explores artificial intelligence models, including Automatic Speech Recognition (ASR) and Natural Language Processing. While AI models can improve efficiency in criminal investigations, their successful implementation requires evaluation of legal and technical risks. This paper explores the legal and technical challenges of applying ASR models to investigative interviews in the context of the European Union Artificial Intelligence Act (AIA). The AIA provisions are discussed in the view of domain specific studies for interviews in the Norwegian police, best practices, and empirical analyses in speech recognition in order to provide law enforcement with a practical code of conduct on the techno-legal requirements for the adoption of such models in their work and potential grey areas for further research.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 7 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.