The AI and Law webinars, co-organised by the Council of Europe and the University of Strasbourg (UMR DRES), are regular meetings open to a wide public, public decision-makers, officials from international organisations and national administrations and academics, whose aim is to measure the stakes of subjects at the frontiers of digital techniques and the practice of law.
This webinar focusses on the role artificial intelligence techniques such as facial recognition play in the criminal justice system. Deep learning techniques of this type promise to make it considerably easier to identify people from pictures and can be a real boon to police departments, provided the identification is technically reliable and fair – and the person in question was already a suspect before they were identified.
Currently a hot topic, video surveillance coupled with a facial recognition system is gaining traction in Europe while several North American cities have not only stopped using facial recognition, some have prohibited it. Does video surveillance create suspects? Is the fact that someone is present in a specific place – predetermined by the police – constitute reasonable grounds on which to launch a criminal investigation? Is using such combined surveillance and recognition systems a fair and effective way to fight crime?
Clear regulations have been adopted in recent decades to govern the use of investigative measures based on specific techniques such as telecommunications interception, using GPS to track vehicles, and installing surveillance cameras in homes. In Europe, the use of such techniques is reviewed by the European Court of Human Rights. Should similar rules apply when facial recognition is used to identify people?
Turning the mere identification of someone into grounds to investigate or proof of a crime is a legally delicate endeavor. Identification by a witness is governed by rules of criminal procedure. Witnesses must testify and be questioned by defense counsel before a judge, and in some cases their testimony cannot be admitted as evidence. Does AI testify? Does it provide an expert opinion? Can defense counsel question it, or impugn its reliability? Do otherwise benign observations produced by AI, such as that a driverless car was on the road, play a role in criminal proceedings?
- Thomas Lampert, Ph.D., Chair of Artificial Intelligence and Data Science, University of Strasbourg
- Kate Robertson, Criminal and regulatory litigator, Markson Law, Toronto
- Sabine Gless, Ph.D., Professor of Criminal Law and Procedure, University of Basel
Hosted by Juliette Lelieur (University of Strasbourg) and Yannick Meneceur (Council of Europe)