The FBI used facial recognition software to track down a man who was involved in the storming of the US Capitol in January.

Through public footage, Stephen Randolph was recognized by the US investigation agency on his girlfriend's Instagram page,

The Huffington Post

writes on


Randolph had been wanted by security since January.

He is suspected of assaulting two officers, one of which was captured on video.

The FBI pulled multiple photos of the riots that Randolph was featured on through facial recognition software.

The software found a face match on Randolph's girlfriend's public Instagram page, where he was featured in one of the photos.

Via that route, the FBI came to the Facebook page of the suspect.

On April 13, two undercover agents came to Randolph, who admitted the videotaped assault and his involvement in the riot.

He was arrested by the FBI on Tuesday.

It is not known which software the security service used.

The FBI does write in a document that the software "has shown in the past that it is very reliable" and that it is open source software.

This means that the user can tinker with the source code himself.

Criticism of the use of facial recognition software

Using facial recognition software to track down criminals and suspects is not without its drawbacks.

It happens that the software does not work optimally and points to the wrong person.

This happened last year in the American city of Detroit.

There, a man was wrongly arrested because the software mistook him for a criminal.

In the Netherlands, the police also use facial recognition software to track down criminals, called CATCH. That system is critical, because a database contains tens of thousands of photos of people, of whom it is not always known whether they are correct.