Policy Observatory
In a letter to Senators Grassley and Leahy, EPIC has urged the Senate Judiciary Committee to investigate the FBI's "Next Generation Identification" program. NGI is the most extensive biometric database in the world and raises many privacy risks. In a recent FOIA case, EPIC v. FBI, EPIC obtained documents which show that the FBI accepted a 20% error rate for facial recognition matches. EPIC and over 30 organizations have urged Attorney General Holder to conduct a privacy assessment of NGI, but the program has since gone fully operational without the required evaluation.
Federal and local law enforcement across the US are adopting sophisticated facial recognition technologies to identify citizens on the streets and in social media by matching faces to massive databases.
At the end of August 2017, German police has been testing a facial recognition software at Südkreuz train station in Berlin. The system was tested on 300 volunteers. The goal was to evaluate the accuracy of the software in recognising and distinguishing them from the crowd – a feature that the police hopes to ultimately use to track and arrest crime and terrorism suspects.
Facebook announced that it will start using its facial recognition technology to find photos of you across its site, even if you aren't tagged in those photos. The idea is to give you more control over your identity online by informing you when your face appears in photos, even those you don't know about. According to a Facebook blog post, the new feature is powered by the same AI technology used to suggest friends you may want to tag in your own uploaded images.
The Dutch police has a facial recognition database containing "well over a million" faces of people taken into temporary detention (and consequently in the "strafrechtdatabase"). According to the news source, police ran approximately one thousand suspects through it, which led to 93 hits. There are at least three humans involved in the process as safeguarding measure: one before a photograph is fed through the algorithm to look at consider aspects such as the quality of the picture. Two humans are checking the results of the automated process. Furthermore, a hit is seen as in "indication" of someone's identity, not as definite proof.
In the Netherlands the use of facial recognition technologies for marketing and profiling purposes of customers seems to be increasing. According to the source, this is legally allowed in cases where costumers provide consent. The article highlights the mainstreaming of biometrical technologies. It is questionable if the general public agrees to the widespread use of such technologies and what the impact on societal interaction will be once the technologies have become ubiquitous.