Berlin Südkreuz rail station algorithmic surveillance
Released: August 2017
Berlin Südkreuz is a rail station and transport interchange junction in southern Berlin that has been used by a number of recent German governments as a laboratory for live biometric and other forms of surveillance.
Facial recognition pilot
In August 2017, Germany's Ministry of the Interior started a six-month pilot, later extended to twelve months, to assess the facial recognition capabilities of three systems tracking 312 volunteers wearing transponders and who were added to a special police database.
The test elicited a strongly critical reaction from civil and privacy rights groups for its potential impact on privacy and other fundamental rights. It was also criticised for the apparently low quality of its outputs.
Interim results of the pilot published in December 2017 indicated that 84.7% of people were correctly identified by the three systems, a figure contested by activists. The test was described by Florian Gallwitz, a facial recognition expert at the Nuremberg Institute of Technology, as 'a clear failure'.
Suspicious behaviour monitoring
In June 2019, the authorities started a pilot at Berlin-Südkreuz to test algorithms supplied by IBM, Hitachi, Funkwerk and G2K Group to detect suspicious behaviour focused on six scenarios, including unattended luggage, acts of violence, people lying down or entering blocked areas such as construction sites.
The project again involved volunteers who were asked to do things to attract the attention of the systems. The pilot provoked another round of negative media coverage.
Operator: Bundespolizei (BPOL); Deutsche Bahn
Developer: Dell/Herta Security; AnyVision; IDEMIA; IBM; Hitachi; Funkwerk; G2K Group
Sector: Govt - transport
Purpose: Strengthen law enforcement
Technology: Behavioural analysis; CCTV; Computer vision; Facial recognition; Object recognition; Neural network; Deep learning; Machine learning
Issue: Accuracy/reliability; Privacy; Surveillance
Transparency: Governance; Marketing
Fontes C., Hohma E., Corrigan C.C., Lütge C. (2022). AI-powered public surveillance systems: why we (might) need them and how we want them
Report for the Greens/EFA in the European Parliament (2021). Biometric and Mass Surveillance in EU Member States
Eireiner A., V. (2020). Imminent dystopia? Media coverage of algorithmic surveillance at Berlin-Südkreuz
Bundespolizeipräsidium Potsdam (2018). Biometrische Gesichtserkennung (pdf)
Deutscher Bundestag (2018). Answers to Written Questions
News, commentary, analysis
Published: February 2023