Rail providers push back on false info about facial recognition
Facial recognition will not be used, and personal data not recorded by new customer frequency measurement systems installed in rail stations, according to rail providers SBB GmbH and Swiss Rail.
The two firms pushed back against what they called false information spread by media indicating that the systems would use face biometrics to collect personal identity data, putting privacy at risk. Both SBB and Swiss Rail said computer vision technology will be used to record the flow of commuters more accurately.
“We don’t need face recognition,” says Alexis Leuthold, head of Real Estate Management for SBB, in a statement. “We want to know how the subgroups of our customers behave in the train station, for example people with prams, travelers with skis or bicycles. We don’t need to know who each person is or what their name is. That is none of our business and is simply irrelevant.”
Leuthold said the tender documents for the project “expressly require that the providers have to ensure that they comply with the data protection law and describe how they do this.” SBB has gone so far as to issue a specification and clarification of the tender, to address “misunderstandings and misinterpretations.”
Still, critics have pointed to language in the original tender that says a person should be able to be “clearly recognized” while in the station, including traits such as gender and age — which implies the collection of biometric data via face scans.
Likewise, consumer advocates in Switzerland have pointed out flaws in Swiss Rail’s claims that their surveillance cameras will not be used to collect personal ID information, according to a report in Le News. The company says it will not identify passengers, but instead use the system to improve placement of sales points, restaurants and other amenities. However, Angela Müller from the tech watchdog AlgorithmWatch pointed out that, even if the company says it will not identify passengers, the infrastructure will allow for facial recognition.
