Icon: enlarge
Well documented: The rioters filmed themselves and their friends with their mobile phone cameras
Photo: Ken Cedeno / imago images / UPI Photo
Since Trump supporters broke into the U.S. Capitol on January 6 and rioted there, the FBI has been looking for those involved.
After just a week, according to FBI director Christopher Wray, investigations had started in 200 cases and 100 people were arrested.
Apparently tens of thousands followed the call to provide the authorities with information about the violent riots.
To date, the FBI has received more than 140,000 videos and photos with leads.
A student from the greater Washington DC area now goes one step further.
He evaluated the 827 videos that were recorded during the riots in and around the Capitol and uploaded to the right-wing social network Parler.
The result is a website that features 6,000 portrait photos of all of the people who appear in those videos in some way.
This was made possible because activists said they had downloaded Parler's entire database before the network went offline.
Amazon, on whose servers the network was running until then, had ended the collaboration.
In addition, the system was apparently so badly secured that it was no problem for the activists to even back up Parler posts that were deleted by the users.
Face recognition for everyone
The man told the US magazine "Wired" that he extracted the individual faces automatically from the video material using freely available facial recognition software and open source algorithms for machine learning.
He wanted to make it possible for everyone to see whether the people in the picture might include people he or she knew.
There is a link to the FBI's tipster page at the top above its picture overview.
It is "quite possible that many people who have been on this website will now have to face consequences for their actions in real life," said the student, who does not want to reveal his identity, the magazine.
On his Twitter account, the anonymous activist reports on attempts to take the website out of the network with a so-called overload attack (DDoS).
In addition, strangers tried to take over his account with the provider who hosts the website.
However, these attempts are doomed to failure, he writes to the attackers, because "Two-factor authentication is activated ... good luck, you will need it."
What the software cannot do, the user should do
The anonymous activist also deals with the personal rights of the people shown just as carelessly as with the attacks on his website.
He left the pre-sorting of the images and the filtering out of duplicates to the software.
A review of the extent to which the people shown actually participated in the violent protests obviously did not take place.
This task should now be taken over by the users of the site.
Just as he relies on clues from outside to find information about the Capitol striker, he also leaves the sorting out of bystanders to others.
He thanked us several times via Twitter for references to pictures of people who did not take part in the riots and promised to remove them from the flood of images.
Evan Greer of the civil rights organization Fight for the Future warns Wired about this approach and about the dangers that the use of facial recognition technologies in his organization generally harbors: “Regardless of whether it is used by an individual or by the government, this technology has profound implications for human rights and freedom of expression. "
The developer of the project counteracts this by counteracting possible misuse of his data pool by neither providing a search function nor giving the opportunity to automatically compare the photos shown with images from other sources.
Despite all the criticism, he hopes that his efforts will bring "tangible results," that is, indications of further accomplices to the FBI.
Icon: The mirror
