Facial Recognition Software being adapted and utilized by the FBI

Singularity University updated their website with an article, THE FBI’S MASSIVE FACIAL RECOGNITION DATABASE RAISES CONCERN yesterday. The article talks about FBI beginning to use the facial recognition database with facial recognition software, even though the algorithm written for it continues to produce false positives and false negatives very frequently. FBI stance is that basically they will continue to use it anyways according to papers obtained by the Electronic Frontier Foundation in a FOIA (Freedom of information act) lawsuit. Right now there are 16 million images mostly from non-criminal records like driver’s licenses; however, the number is expected to hit 52 million face images by 2015. The FBI’s criterion for accuracy is that the search should give out top 50 top images about 85% of the time! Furthermore, the FBI is going to obtain about 200,000 images from repositories, not mentioning what these repositories might be. The law enforcement agencies will also drop in 700,000 images, while the images on social media are off-limit.
As the database grows, the number of false positives will increase with it, and some innocent person might end up in the FBI’s interrogation room. Another ethical question arises by using the driver license data in this process; these people will be presumed guilty until proven innocent. How responsible is this action, and how to use this technology responsibly, and ethically?

2 Responses to Facial Recognition Software being adapted and utilized by the FBI

  1. In my opinion, the facial recognition software, if used correctly, could become a very useful tool for law enforcement. However, the fact that it gives false positives raised concerns for many people. I completely agree with Khan’s point that people will be presumed guilty until proven innocent. The problem with this is, some people may stay guilty due to many factors surrounding the case. Chris Owen from “The Telegraph” boldly compared facial recognition law enforcement to witch hunting. It sure does sound like witch hunting, when people are being accused of a crime because they “look” like the person described by the footage.
    All the concerns surrounding facial recognition software should encourage law enforcement bureaus to take action. Maybe the accuracy of the software could be improved somehow. If the accuracy increases, some might reconsider their stand on this case. Now the question is, is it possible to develop a bulletproof facial recognition software, that gets it right 100% of the time? If the answer is no, the debate surrounding facial recognition software and database will continue to take place for a long time.

  2. As Chris said, I do believe as well that if something is used correctly, it could prove useful to the general public by use of law enforcement or by any other general use. The fact that this software generates false positives almost half the time is really striking though. If the FBI’s stance on the topic is that they will continue to use a software that has proved to be wrong on multiple occasions really worries me and it should worry the general public as well. When a piece of software is extremely accurate, it almost becomes a liability. In this case, the liability aspect comes into play as someone is using the extremely accurate piece of software to be target people who may be responsible for committing a crime. The person targeted is suddenly guilty and has nothing that he can use to get rid of the blame. If there was an 100% method of always being right when targeting criminals, there would obviously be no worries in the system, but currently in our technological state I believe that there is no 100% way of being sure of anything.