In the American city of Detroit, a black man was wrongfully arrested because facial recognition technology confused him with a criminal. The American human rights organization ACLU reports this on Wednesday in a complaint to the Detroit police. The use of facial recognition has been under fire for some time because the technology is said to recognize black people less well.

The man, Robert Williams, was detained at the police station for a day because police facial recognition software confused his face with that of another black man who stole five expensive watches in 2018. As far as is known, this is the first time that face recognition technology has caused a false arrest.

The facial recognition software had run the thief's face through a database of 49 million faces. That's where Williams's face came out as a match. After sleeping overnight at the police station, the officers saw that Williams's appearance did not match the thief's appearance and was allowed to return home. According to Williams, the agents told him that the computer was "wrong."

The ACLU calls on the Detroit police to stop using facial recognition "as this case proves that both the technology is flawed and investigators are not able to use such technology".

Facial recognition under attack

The use of facial recognition has been under attack for some time. For example, critics say that the large-scale recognition of faces violates civil rights and privacy. The technique is also less able to recognize women and black people.

Earlier this month, Amazon announced it would stop supplying facial recognition software to police in the United States for a year. Microsoft also says it will not provide the police with the technology. The companies want legislation to be introduced so that the use of the technology can be regulated.

IBM previously announced that it would completely discontinue the development of facial recognition technology. The company says the technology is being used for mass surveillance and ethnic profiling, among other things.


This is how facial recognition works for the police