One of the most controversial aspects of “policing in the 21st century” is the notion that if police can sweep all information into centralized databases and let artificial intelligence do the investigative work of making connections, then crimes can be prevented before they happen: pre-crime.
One would imagine that if police felt that the public would wholeheartedly embrace this technique, they would be quick to publicly disclose its use and effectiveness. Troublingly, this has not been the case. It has taken the intrepid work of investigative reporters, as well as activists dedicated to stopping warrantless police surveillance, to uncover the details that have been mired in secrecy at several high-profile police departments around the country.
Perhaps the most disturbing of the programs came to light in February by Ali Winston who wrote an investigative report for The Verge exposing a contract between the New Orleans PD and Palantir Technologies. The program was so secretive that not even the city’s elected representatives were made aware of it until the article was published. Subsequently, New Orleans did cancel the contract with Palantir. However, as I reported at the time, Palantir is one of the foremost leaders in the field of predictive algorithms for intelligence work internationally and is directly connected to the CIA. One would naturally have to posit that other police departments must be pursuing similar relationships with Palantir.
A new public records request was filed by activists for the Stop LA Spying Coalition that did, in fact, uncover Palantir’s presence at the LAPD. The results were provided to In Justice Today. The picture that the documents paint looks similar to that which has been detailed in New Orleans and offers additional insight about what is already being utilized, apparently without citizen input and without proper oversight. The result is a “probable offender” list compiled by computer algorithm that has put a target on the backs of anyone swept up in the data stream.
These surveillance reports identify “probable offenders” in select neighborhoods, based on an LAPD point-based predictive policing formula. Analysts find information for their reports using Palantir software, which culls data from police records, including field interview cards and arrest reports, according to an updated LAPD checklist formula, which uses broader criteria than the past risk formula the department was known to have used. These reports, known as Chronic Offender Bulletins, predate Palantir’s involvement with the LAPD, but since the LAPD began using the company’s data-mining software in September 2011, the department claims that bulletins that would have taken an hour to compile now take “about five minutes.”
Los Angeles police argue that targeting “chronic offenders” in this manner helps lower crime rates while being minimally invasive. But the Stop LAPD Spying Coalition, a community-based alliance that has advocated against increased LAPD surveillance efforts since 2012, paints a different picture of the Chronic Offender Bulletin program. The group calls it a “racist feedback loop” in which police surveil a set number of people based on data that’s generated by their own racially biased policing, creating more monitoring and thereby more arrests.
Field interview cards, for example, which provide information for the predictive checklist, often result from on-the-street racial profiling, argues Jamie Garcia, the lead organizer on the Predictive Policing campaign with the Stop LAPD Spying Coalition. “When we look at LAPD stops, the black population is completely overrepresented,” said Garcia in a phone call. The directives, she says, direct officers “to find these people and to basically harass them…. If you’re constantly being surveilled, constantly being harassed, the chance of something going wrong … The next thing you know, you’re a chalk outline.”
Legal scholars have noted that the institutionalization of risk formulas like the LAPD’s Chronic Offender program checklist can exacerbate existing patterns of discrimination by oversampling those already discriminated against, generating even more biased data that justifies further discrimination.
The LAPD declined IJT’s request for interview, and did not provide answers to written queries about the program. In an email to IJT, Palantir spokesperson Lisa Gordon confirmed that Palantir is used in the creation of Chronic Offender Bulletins, but stressed that the software does not automatically generate the reports and that the selection and vetting of people on these lists are part of “a human-driven process.”
How Pre-Crime Investigations Begin
The target identification process starts with a LAPD analyst looking for “probable offenders” by surveying police records. According to the LAPD documents, analysts deploy Palantir’s file-organizing software to conduct “work-ups” of these individuals, looking for records that add points to their predictive risk scores, which are based on factors, such as whether they are on parole and their number of police contacts in the last two years. Below is an image of one of these “work-ups,” generated a few months before the department adopted Palantir, which IJT found online, completely unredacted, in a May 2013 LAPD presentation.
Once again, we see that not only have the details been kept secret until a forced reveal, but any ensuing questions from the public subjected to the surveillance are treated with disdain. This does nothing to alleviate concerns that these programs have a nefarious purpose. Moreover, the debate is still out as to whether this technology is accurate enough to be used in law enforcement. Even experts in the field have serious concerns (Please see: “Predictive Algorithms Are No Better At Telling The Future Than A Crystal Ball”).
Every police department should be open to requests from the public to explain their conduct. The fact that they continue to shirk their responsibility only highlights why this issue is so important for freedom. Let’s hope that the continuing efforts to keep police accountable will have the same result in Los Angeles as those efforts had in New Orleans.
Image credit: Phys.org