When Machines Become Witnesses: How is AI Changing Court Proceedings?

Podcast with Sabine Gless

Smartwatches, Smartphones, car assistants — digital companions are with us around the clock. They measure, store, and observe us, often without our awareness. But what happens when the data they collect suddenly appears in court? In the latest episode of the Max Planck podcast “Ach Mensch!”, Max Planck Research Fellow Sabine Gless explores how smart technology and artificial intelligence are transforming criminal proceedings.

At first glance, data from smart devices might seem like objective, neutral evidence. Yet, as Gless explains, things are rarely that simple. A car’s fatigue warning system, for instance, cannot recognize an animal crossing the road. It only registers a sudden steering movement, and might mistakenly interpret it as a sign of microsleep.

Another major challenge is transparency. The algorithms behind AI systems are rarely fully accessible. Manufacturers often keep their code secret to protect intellectual property, and the law allows them to do so. This means it often remains unclear how an AI system actually arrives at its conclusions.

If artificial intelligence is to be used in criminal trials, procedural safeguards would have to evolve. In Europe, every accused person has the right to effectively challenge incriminating evidence. With AI-generated evidence, however, this becomes difficult: algorithms cannot be questioned or cross-examined like human witnesses. New legal provisions would be needed in Germany, such as specific rules on what might be called “AI evidence”.

Listen to the full podcast episode (in German): www.mpg.de/podcasts/ach-mensch

Learn more about Sabine Gless’s research group “Algorithmic Profiling and Automated Decision-Making in Criminal Justice”: csl.mpg.de/en/sabine-gless