Berlin startet KI-Überwachung am Kottbusser Tor
30 Kameras sollen ab November automatisch Gewalttaten und Notfälle erkennen. Ganz so anonym, wie zunächst dargestellt, ist die Überwachung allerdings nicht.
The Berlin government plans to launch AI-powered video surveillance at Kottbusser Tor beginning in November. Thirty cameras will monitor the area around the clock, automatically detecting possible violent incidents or emergencies. After a test run, the system is set to go live at the start of December. Kottbusser Tor is one of seven so-called crime hotspots in Berlin.
The legal basis for the AI deployment was created by Schwarz-Rot with the reform of the General Security and Order Act. Iris Spranger, the Interior Minister, cited the high crime rate in the Kottbusser Tor area for the move, noting that over 900 criminal acts, including more than 300 physical assaults, had been registered in the past 12 months.
"We want to detect dangers or problematic situations earlier, react quickly, and provide people with protection," Spranger explained. The police will demonstrate how the system works, with statisticians staging a confrontation in front of the cameras while the system issues an alarm inside the police station. The video feed appears on a black monitor, and a police officer decides whether to intervene.
The AI analyzes the video material continuously for certain movement patterns, simplifying the portrayed persons into a kind of "stick figure" and evaluating their movements. If the system considers a situation suspicious, it will sound the alarm. The "stick figure" representation applies only to the AI's analysis; upon receiving an alarm, officials view the normal video feed showing the persons.
The Interior Minister acknowledged that her previous statement about the images remaining anonymous was incorrect, as officers need "the maximum amount of information" to make their decision. The video recordings will be stored for 28 days and can be used for criminal investigations if they show a criminal act. The police claim the data remains under government control and is stored in a secure Berlin Police data center, encrypted and compliant with BSI guidelines.
The AI's accuracy in the real world, however, remains uncertain. Hamburg's experience shows that 99% of AI-triggered situations were false alarms. The Police Director expects significantly better results in Berlin, stating that they consciously chose a different system. "We are very convinced of the selection we made." The future of the AI project depends on the political landscape.
After a possible government change to Rot-Rot-Grün, the AI surveillance could become a point of contention. The left-wing group had previously criticized the use of surveillance cameras in public spaces. Spranger emphasized, however, that she wants to maintain the project. "Even Greens and the Left must ensure safety in Berlin."
The question of what happens to the AI cameras then is not yet resolved. In a proposal paper, SPD, Greens, and the Left agreed to evaluate the video surveillance. The fate of the AI cameras will likely depend on the outcome of the coalition negotiations.
Written by urgent.news from Politico EU's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.