Cities around the United States are coming under pressure to scrutinise policing technology
By Avi Asher-Schapiro
NEW YORK, June 17 (Thomson Reuters Foundation) - As pressure mounts to address police brutality and racism, California's Santa Cruz is poised to become the first U.S. city to ban predictive policing - despite headquartering the firm that pioneered the technology.
Used by police forces across the United States for almost a decade, PredPol Inc - short for predictive policing - relies on an algorithm to analyse police records and identify crime-ridden areas to determine when and where officers patrol.
But critics says it reinforces racist patterns of policing - low-income, ethnic minority neighbourhoods have historically been overpoliced so the data shows them as crime hotspots, leading to the deployment of more police to those areas.
"We have technology that could target people of colour in our community - it's technology that we don't need," Justin Cummings, the seaside city's first male African-American mayor, who proposed the ban last year, told the Thomson Reuters Foundation.
"This is something that targets people who are like me," said Cummings, who made headlines this month for kneeling beside Santa Cruz police chief Andy Mills to protest the death of George Floyd in police custody, which sparked global outrage.
PredPol did not immediately respond to requests for comment.
On its website, PredPol said that its technology helps police fight crime and that Santa Cruz police reported a 19% reduction in burglaries since implementing its programme while Los Angeles Police saw a 25% fall.
"PredPol's software technology does not pose any personal privacy or profiling concerns," the company said on its website.
"PredPol does not collect, upload, analyse or in any way involve any information about individuals or populations and their characteristics."
Across the United States, changes are being made to curb police violence and racial injustice, including bans on neck holds and "no-knock" warrants to forcibly enter homes and calls to defund police departments and for the use of body cameras. A number of technology firms, including Amazon and IBM, said this month they would limit police use of facial recognition software, after a two-year battle with civil liberties activists concerned that inaccurate matches could lead to unjust arrests.
Cummings said he plans soon to sign into law a ban on predictive policing and facial recognition technology - which can misidentify people with darker skin, raising fears of harassment - passed in March by a Santa Cruz council committee.
"It's high time we really step back and honestly assess the impact policing technology has on our communities," he said.
Santa Cruz's move could prompt other cities to follow suit amid growing criticism of predictive policing's racial bias, said activists working to stop law enforcement using it.
About 35 miles (56 km) south of Silicon Valley, the Santa Cruz Police Department rolled out PredPol in 2011, becoming one of the first forces in the nation to use advanced predictive algorithms to fight crime.
"Over the years, Santa Cruz became closely associated with this technology, and played a major role exporting it around the country, which is part of why the mayor's legislation is so significant," said Matt Cagle, a lawyer with the ACLU.
Critics of predictive policing say Santa Cruz has been central its expansion, piloting PredPol's software and being used in the firm's marketing materials, while ex-city officials have gone on to join PredPol and pitch its technology elsewhere.
Police spokesman Zach Friend, who helped test PredPol in Santa Cruz, went on to serve on its board, and former mayor Ryan Coonerty worked as PredPol's director for government relations.
Other cities, including Los Angeles, Chicago and New Orleans, have rolled back their use of predictive policing, but no other city has gone as far as Santa Cruz and explicitly banned it, said Cagle, who is based in Northern California.
"One important response to current calls for police reform is to ban dangerous and racially discriminatory surveillance technology, like predictive policing," he said.
Although PredPol does not say which police forces it has contracts with, its website says the technology is being used to help protect one in 33 people in the United States.
Critics say PredPol's software encourages punitive overpolicing, by deploying officers to arrest people for low-level property crimes like burglary, theft and shoplifting.
"It's propagating racism," said Cathy O'Neil, an award-winning mathematician who audits algorithms for bias.
Using drug crime data from Oakland, California, PredPol's algorithm would deploy police to ethnic minority neighbourhoods at double the rate of white neighbourhoods, a 2016 study in the Royal Statistical Society's Significance magazine found.
As crime data is generated by the police, it does not accurately represent where crime actually occurs and predictive policing simply reproduces those policing patterns, O'Neil said.
"They want to confuse people by using math - and make it seem like if you aren't a data scientist you won't be able to understand it," said Jamie Garcia, an organiser with Stop LAPD Spying Coalition.
The activist group campaigned from 2016 for Los Angeles to stop using PredPol. In April, police in the city - where Rodney King's death led to riots in 1992 - said they had abandoned the tool due to financial constraints caused by the new coronavirus.
Santa Cruz police chief Mills, who took over the department in 2017, said he stopped using PredPol in the first six months on the job and he had not seen a significant change in crime rates as a result.
"I don't want to be in a place where we are targeting people of colour based on a computer system," he said.
(Reporting by Avi Asher-Schapiro; Editing by Katy Migiro. Please credit the Thomson Reuters Foundation, the charitable arm of Thomson Reuters, that covers the lives of people around the world who struggle to live freely or fairly. Visit http://news.trust.org)
Our Standards: The Thomson Reuters Trust Principles.