The Minister Wants Your Face
Thomas de Maizière stood on the platform at Berlin’s Südkreuz station in August 2017 and watched a camera scan faces in a crowd. He was, by all accounts, delighted. If this worked,
Germany’s Interior Minister told reporters, it would be an incredible security gain for the population of the Federal Republic.
He could envision rolling out biometric facial recognition nationwide—across all major transport hubs—within roughly a year, pending the right legal framework. The terrorist attack on Berlin’s Breitscheidplatz Christmas market in December 2016 had killed twelve people, and de Maizière was not wrong that something terrible had slipped through the net. The cameras, in his mind, were the net.
The pilot project at Südkreuz had been running since earlier that summer: volunteers enrolled their faces in a database, and the system was tested on its ability to identify them in a moving crowd. De Maizière dismissed the privacy objections with the ease of a man who hadn’t felt the force of them. A camera already records people, he argued—temporarily, without identifying anyone. This system only checks whether your face matches a name on a wanted list, discards the no-match result in seconds, and stores an image only on a hit. What, exactly, was the problem?
Constanze Kurz at Netzpolitik.org covered his visit with the careful alarm it warranted. Privacy advocates, the Pirate Party, and organizations like Aktion Freiheit statt Angst (Freedom Not Fear) had already demonstrated at the station and were calling for the project to be halted. Their concern isn’t specifically about this pilot—it’s about the infrastructure being assembled around it and the political precedent of normalizing continuous biometric scanning of civilian transit.
The version of de Maizière’s argument I find hardest to dismiss goes roughly like this: if a system genuinely flags only people on active criminal watch lists, and genuinely discards everyone else within seconds, how is that categorically different from a police officer who recognizes the face of a wanted man? The problem is that "genuinely" is doing a lot of work in that sentence. The history of surveillance infrastructure is almost without exception a history of scope creep. Tools built for terrorism get applied to drug offenses, then to theft, then to whatever a government finds convenient. The technical capability doesn’t shrink. It accumulates.
There’s also the question of what "only" means over time. Only matching against wanted lists—until the lists grow. Only storing a match—until storage policies change. Only deployed at major stations—until that definition expands. De Maizière’s breezy certainty that this isn’t a step toward a police state reads less like reassurance and more like a man who hasn’t seriously sat with the possibility that it might be.
The freedom-versus-security framing is itself a trap, because it implies the trade is clean and proportional. Give up a little privacy, receive a measurable amount of safety. It doesn’t work like that. The surveillance apparatus you build to catch terrorists is the surveillance apparatus available to whoever runs the government next. That’s not a conspiracy theory—it’s institutional behavior. And once it’s everywhere, once every face at every transit hub is being scanned as a matter of course, you don’t reverse it cheaply. You probably don’t reverse it at all.
I’m genuinely disturbed by mass-casualty attacks and I understand the political pressure that generates proposals like this one. De Maizière isn’t a monster. He’s a man with a tool that looks like a solution and a very recent atrocity to point at. That’s exactly the kind of moment when the worst decisions get made with the best intentions.