Baden-Wurttemberg Coalition Plans CCTV Combining Facial Recognition and Behaviour Analysis

The new Greens-CDU coalition agreement in the German state of Baden-Wurttemberg includes plans for public video surveillance that combines live facial recognition with behaviour analysis, a step that would make the state the first in Germany to operate cameras running both technologies at the same time.

According to the coalition agreement, the cameras would analyse the behaviour of people in view while simultaneously measuring their faces, using facial recognition to determine whether a person is of interest to police. The coalition document also references detecting dangerous objects and enabling biometric remote identification, and it states an intention to create a legal basis for automatically comparing camera images against images that are publicly available on the internet.

A pilot is planned for Mannheim and two further cities that have not yet been named, for three locations in total. Mannheim has run behaviour-analysis cameras for roughly eight years, and the coalition plan would add facial recognition to that existing deployment rather than starting from scratch. The agreement also signals an intent to remove rules that currently limit such cameras to designated higher-crime areas, which would allow wider installation across public spaces.

The plan drew a response from the state’s data protection authority. Baden-Wurttemberg’s State Commissioner for Data Protection and Freedom of Information, Professor Tobias Keber, issued a public statement on the newly published coalition agreement, raising concerns about the surveillance provisions. Digital rights coverage of the agreement, including by the German outlet netzpolitik.org, characterized the proposal as cameras that assess who a person is and how they behave.

Baden-Wurttemberg would be acting within a wider German pattern. Facial recognition cameras are already used in Frankfurt, and other states, including Hamburg, have adopted or are developing legal frameworks for real-time remote identification and for training systems on footage of bystanders. Federal lawmakers have separately pursued changes that would let police match faces against publicly accessible online images.

Such deployments intersect with the European Union’s AI Act, which tightly restricts real-time remote biometric identification in public spaces and bars untargeted scraping of facial images. The European Parliament has voted to delay the AI Act’s high-risk compliance deadlines to December 2027, a category that covers biometric remote identification systems, leaving the legal boundaries for state-level programs a continuing point of contention.

Sources: netzpolitik.org, IamExpat

–

By the ID Tech Editorial Team