False Arrest Highlights Issue of Racial Bias in Facial Recognition

A Michigan man has teamed up with the ACLU to file a complaint against the Detroit Police Department in what is currently the first known instance of an arrest based solely on a false facial recognition match. The complaint is asking the department to end its use of facial recognition software in its investigations.

The arrest in question occurred in January of 2020, when police took Robert Williams into custody and held him for nearly 30 hours after accusing him of stealing $4,000 worth of watches from a retail store. The only evidence against him was a facial recognition match that linked blurry surveillance footage of the robbery to Williams’ driver’s license photo.

The problem is that Williams is not the person depicted in the video, so the false match only served to reinforce the racial bias that has been observed in many facial recognition systems (Williams is a 42-year old Black man). It also highlighted the severe shortcomings of the police procedures. Most facial recognition policies – including those of the Michigan State Police – stress that a facial recognition match is not sufficient for an arrest, but should instead only be a lead that prompts a more thorough investigation.

However, that does not seem to have happened in Williams’ case. Had they investigated the matter, the police would have discovered that Williams was 40 minutes away at the time of the robbery. Instead, they only presented a security guard who had not witnessed the robbery (and had only seen the security footage) with a lineup that included Williams’ license, and then decided that was sufficient grounds for an arrest.

The ACLU notes that while Williams was released, the incident will still have long term repercussions. Williams was arrested in front of his entire family, and his DNA, fingerprints, and personal information have now been entered into a law enforcement system that frequently uses deadly force against Black individuals.  

“I never thought I’d have to explain to my daughters why daddy got arrested,” said Williams. “How does one explain to two little girls that a computer got it wrong, but the police listened to it anyway?”

The ACLU went on to criticize the police for their lack of transparency, arguing that law enforcement will often not disclose when facial recognition was used to identify a suspect. That’s why the organization believes that false matches have likely led to other arrests. Williams is simply the first to receive any kind of publicity.

In all likelihood, the incident will give more urgency to the recent calls for police reform. Critics have argued that mass surveillance is an inherent threat to people’s civil liberties, and that the police, in particular, should not have access to such powerful tools if they cannot be trusted to use them responsibly. Several cities have already banned the police use of facial recognition, and major corporations like IBM, Microsoft, and Amazon have announced that they will no longer be selling their technology to law enforcement agencies.

Sources: ACLU, Wired

–

June 25, 2020 – by Eric Weiss