Capitol riot suspect caught after facial recognition software found his girlfriend’s Instagram
A suspected US Capitol rioter accused of attacking three cops and whose image was circulated on social media with the hashtag #GrayCarharttHat was arrested after the FBI used facial recognition software to find his girlfriend’s Instagram.
Stephen Chase Randolph, of Harrodsburg in Kentucky, was arrested on Tuesday and charged with assaulting a federal officer, obstruction of law enforcement during civil disorder and obstruction of justice and Congress, according to the FBI statement.
Randolph was identified as an alleged rioter thanks to sophisticated facial recognition software that matched images of him at the riots to photos uploaded to his partner’s social media from December 2020, the FBI revealed.
Some of the Instagram snaps are said to have shown him wearing the same clothes as the man pictured rioting.
Using open-source software to find a photo of Randolph on his girlfriend’s social media, agents were then able to find Randolph’s own Facebook page and mount an undercover sting.
Agents have yet to disclose further details about the software. But it likely used an algorithm to study the suspect’s facial features, before turning them into a mathematical code. That code is then compared to millions of photographs on databases – or social media – to try and find a face whose features match.
‘Open source’ is a category of software that is accessible to the public for use and can be shared, modified, or enhanced thanks to its available source code which can be altered by programmers.
Federal authorities allege that Stephen Chase Randolph of Harrodsburg, Kentucky (above) took part in the January 6 MAGA riot at the United States Capitol
The use of facial recognition technology has sparked concerns from civil liberties groups that it violates people’s privacy.
Dozens of companies now sell the ability to identify people from pictures of their faces.
As the technology has evolved, some municipalities have banned their police departments from using it to identify suspects.
In February, Minneapolis’ city council voted to ban its police department from using facial recognition technology.
Other cities have also done the same, including Portland, Oregon; San Francisco; and Boston.
Randolph’s arrest has become the highest profile acknowledgement by the FBI that they use the software to hunt down suspects.
According to a federal court affidavit, the face recognition tool ‘yielded results associated with the Instagram page of an individual (“Individual-1”) from Kentucky who appeared to be the girlfriend of the SUBJECT.’
‘Individual-1′s Instagram account also contained a photograph of the SUBJECT (see Figure 11) wearing the same grey toboggan with white “Carhartt” embroidered on the front.’
Last week, two FBI agents working undercover managed to speak to Randolph at his workplace. During their conversation, the topic of the Capitol riots came up.
According to court papers, Randolph told the undercover agents that ‘s*** went crazy’ and that ‘it was f*****g fun’ to take part in the insurrection.
‘RANDOLPH opined that the female police officer likely had a concussion because she was curled up in the fetal position after being pushed to the ground,’ according to federal investigators.
He was arrested and taken into custody on Tuesday, with his case the most high profile acknowledgement by the FBI that they have used facial recognition technology to identify suspected Capitol rioters.
The agency said Randolph was being held in federal custody and was expected to have his first appearance before a federal judge on Thursday.
It was unclear whether he had an attorney who could comment for him.
Federal authorities say that Randolph is the man pictured above as he and several others stormed metal barricades just outside the US Capitol on January 6
Randolph is allegedly seen above pushing through a metal barricade manned by a Capitol Police officer. The officer hit her head on a staircase and lost consciousness as a result of the assault, according to federal investigators
The FBI says that is used facial recognition technology to identify Randolph from his girlfriend’s Instagram account. The image above was posted to Instagram by Randolph’s girlfriend in December 2020. Randolph is seen above wearing the same gray Carhartt knit cap
After the riot, a video of a man purported to be Randolph went viral on Instagram.
It shows the individual assaulting a member of the US Capitol Police and ‘engaging in disruptive or disorderly conduct,’ according to court documents first obtained by Huffington Post.
Randolph was alleged to have been among a crowd that gathered to the west of the US Capitol near the Peace Monument in the roundabout that connects Pennsylvania Avenue NW and 1st Street NW.
According to court papers, Randolph and another individual led the crowd past a fence line and toward a second line of metal barricades manned by uniformed Capitol Police officers.
Randolph, who is allegedly seen in photos and video wearing a gray Carhartt toboggan, a black jacket, and a gray turtleneck, ‘can be seen violently pushing and pulling the barricades until the crowd successfully pushed the barricades down on top of the officers,’ according to federal investigators
On social media, a Twitter account with the handle @SeditionHunters circulated the suspect’s image with the hashtag #GrayCarharttHat, a reference to the knit cap that Randolph was allegedly wearing
Federal investigators then used an ‘open source facial comparison tool’ that has been ‘known to provide reliable results in the past.’ According to a federal court affidavit, the face recognition tool ‘yielded results associated with the Instagram page of an individual (“Individual-1”) from Kentucky who appeared to be the girlfriend of the SUBJECT’
Last week, two FBI agents working undercover managed to speak to Randolph at his workplace. During their conversation, the topic of the Capitol riots came up. According to court papers, Randolph told the undercover agents that ‘s*** went crazy’ and that ‘it was f*****g fun’ to take part in the insurrection
Randolph, who is allegedly seen in photos and video wearing a gray Carhartt toboggan, a black jacket, and a gray turtleneck, ‘can be seen violently pushing and pulling the barricades until the crowd successfully pushed the barricades down on top of the officers,’ according to federal investigators.
At one point, a man alleged to be Randolph can be seen knocking over a police officer, ‘causing the officer’s head to hit the stairs behind her, resulting in a loss of consciousness.’
Randolph is then alleged to have ‘continued to assault two other USCP officers by physically pushing, shoving, grabbing, and generally resisting the officers and interfering with their official duties,’ according to federal court papers.
In late January, the FBI asked the public for help in identifying a man designated as #168-AFO.
On social media, a Twitter account with the handle @SeditionHunters circulated the suspect’s image with the hashtag #GrayCarharttHat, a reference to the knit cap that Randolph was allegedly wearing.
The FBI released these photos which allegedly show Randolph just outside of his workplace last month. He appears to be wearing the same gray Carhartt hat
Federal investigators then used an ‘open source facial comparison tool’ that has been ‘known to provide reliable results in the past.’
The FBI did not immediately give additional details about Randolph’s arrest or charges.
At least a dozen other Kentucky residents have been arrested in connection with the January 6 riot in which a mob forced its way into the US Capitol building.
More than 370 people are facing federal charges in the deadly insurrection, which also sent lawmakers into hiding and delayed the certification of President Joe Biden’s election win.
Many of those who have been arrested and charged were tracked down through viral videos that led investigators to identify them on their social media accounts.
HOW DOES FACIAL RECOGNITION TECHNOLOGY ACTUALLY WORK?
At least 117 million Americans have images of their faces in one or more police databases. According to a May 2018 report, the FBI has had access to 412 million facial images for searches
Facial recognition technology (FRT) is available to the wider public thanks to open source software whose code can be modified or enhanced by programmers.
There are dozens of FRT software available on the market today. The technology is used to log in to mobile phones, access biometric data, gain clearance into secure spaces, and even pay for groceries.
In 2014, FRT first appeared when Facebook announced the rollout of DeepFace, a program that was able to determine whether two photographed faces belonged to the same person.
DeepFace was a smashing success having achieved an accuracy rate of 97.25 per cent.
A year later, Google unveiled FaceNet, which bested its rival with an accuracy rate of 99.63 per cent.
The technology has evolved leaps and bounds since then as open source algorithms on most available software today can net an accuracy rate that is even better than FaceNet.
How does FRT work? There are three stages.
First, an algorithm is trained to detect a face in an image.
The technology uses machine learning algorithms to automatically detect human faces from surveillance cameras, social media and other sources and attempts to match the images with a countywide mug shot database
Faces are then mapped by measuring the distance between nodal points, which are placed in several areas including between the eyes, nose, eyebrows, and the other facial features.
These measurements are then compared with other data sets within a database of other photographs, where a match can be found.
The final step involves identifying the photo by attaching a name to the face using information from the database.
At least 117 million Americans have images of their faces in one or more police databases.
According to a May 2018 report, the FBI has had access to 412 million facial images for searches.
The technology has raised concerns by some who worry it could be a dangerous step toward a surveillance state where people’s movements are tracked the moment they leave their homes.
Earlier this year, Minneapolis’ city council voted to bar the local police department from using facial recognition technology.
Minneapolis joined a small, but growing number of cities that have banned the technology, including Los Angeles, Boston and Portland, Oregon.
Law enforcement has defended the technology as an important a tool and that even a grainy image captured on a security camera or social-media account can lead investigators to a suspect who might otherwise have gone undetected.
Sources: Data Driven Investor, Norton
While there has been some pushback against the use of facial recognition technology, one company has seen its technology increasingly adopted by law enforcement.
Clearview AI, one of the leading face recognition software brands, said that after the Capitol riot, law enforcement agencies have come to rely on their technology. Usage of the brand increased by 26 per cent, according to the firm.
A report by the federal Government Accountability Office found that between 2011 and 2019, law enforcement officials around the country have performed 390,186 searches using face recognition technology.
Last summer, as Black Lives Matter protesters took to the streets in the wake of the police-involved death of George Floyd, the usage of face recognition technology spiked.
Face recognition technology is also widely deployed by US Customs and Border Protection personnel that man ports of entry across the country.
In 2020, CBP used face recognition on more than 23 million travelers – an increase from 19 million the year prior.