Cops often hush up use of facial recognition tools – report

Police around the United States are routinely using facial recognition technology to help identify suspects, but those departments rarely disclose they've done so - even to suspects and their lawyers. 

Documents concerning the use and disclosure, of facial recognition technology were provided to the Washington Post as part of its ongoing investigation into use of the technology in the US,  but only from around 40 departments in 15 states out of the "more than 100" departments who were asked. Most, WaPo noted, declined to answer anything. 

Police records reportedly indicate that, aside from not disclosing that facial recognition technology, police also frequently obscured use of the technology by saying they identified suspects "through investigative means," while others have outright policy documents that tell officers to "not document this investigative lead." 

In multiple cases documented in police reports and court filings, WaPo found those charged with crimes based on facial recognition often weren't aware that it had been used to identify them until after they were already in jail – several times incorrectly.

"Misidentification by this type of software played a role in the wrongful arrests of at least seven innocent Americans, six of whom were Black," WaPo said. "Charges were later dismissed against all of them." 

One of those individuals, an Atlanta resident, ended up spending six days in jail for a crime committed in Louisiana - a state he had never visited. 

Elsewhere, records from the Miami Police Department viewed by WaPo indicate officers there conducted 2,500 facial recognition searches in the past four years that have led to more than 186 arrests and over 50 convictions. Of those arrested based on facial recognition results, WaPo found, less than seven percent were informed the technology had identified them. 

Police don't appear to be sharing their use of facial recognition with other local officials, either: Katherine Rundle, state attorney for Miami Dade County, told WaPo that, "before being contacted by The Post, Miami police had not informed her office about their use of facial recognition in the vast majority of cases." 

A tangled web of questionable technology

Facial recognition has well-known biases, and similar technologies that use AI and algorithms to identify people, like remote ID verification technology, have been found to have similar shortcomings, frequently failing to identify people properly. 

But facial recognition is easy - just run a crime scene photo through a massive database of photos scraped from the internet and blam: Suspect "identified." 

• Microsoft doesn't want cops using Azure AI for facial recognition
• Keir Starmer says facial recognition tech is the answer to far-right riots
• FTC bans Rite Aid from using AI facial recognition in stores for 5 years
• Harvard duo hacks Meta Ray-Bans to dox strangers on sight in seconds

The combination of easy-to-use technology with bias and frequent failures (one reviewed by WaPo suggested both basketball legend Michael Jordan and a "cartoon of a black man" as possible matches for a suspect) have led to multiple local governments passing laws to ban the use of facial recognition by police, but that hasn't stopped them.

Earlier reporting, which has been investigating facial recognition use by law enforcement for the past six months, also found that police in localities that ban facial recognition often simply have other police departments conduct the searches for them, side-stepping their own laws.

Clearview AI, whose tool is used by law enforcement agencies throughout the US, has policies in place that prohibit sharing of face recognition data between government agencies, but that doesn't appear to have stopped law enforcement. The company didn't respond to questions for this story. The Register also reached out to law enforcement agencies mentioned in the report, but we haven't heard back from any. 

The report raises a very interesting question about the legality of facial recognition searches when such information is withheld from a US suspect and their legal teams, as appears to have been the case on several occasions: Whether such an action would qualify as a violation of the Brady rule.

That rule, enacted in 1963, requires prosecutors to disclose any evidence that could be favorable to the defendant to their lawyers. Unfortunately, there aren't any federal laws concerning facial recognition, and WaPo notes that courts have come down on both sides with regard to the disclosure question. 

The most recent attempt to federally regulate facial recognition tech, proposed in the Senate last year, would have banned any law enforcement agency from using facial recognition tools that weren't first authorized for such use by Congress. That bill hasn't budged since being sent to committee in March, 2023.

The US government itself even concluded earlier this year that facial recognition technology needed formal regulation due to its potential to violate privacy and civil rights laws, but the country still doesn't have any regulations on its use. ®