Tuesday, December 23, 2014

Cops scan social media to help assess your ‘threat rating’

By Brent Skorup
December 12, 2014
http://blogs.reuters.com/great-debate/2014/12/12/police-data-mining-looks-through-social-media-assigns-you-a-threat-level/
minority-report1
Tom Cruise in Minority Report. Courtesy of 20th Century Fox
A national spotlight is now focused on aggressive law enforcement tactics and the justice system. Today’s professional police forces — where officers in even one-stoplight towns might have body armor and mine-resistant vehicles — already raise concerns.
Yet new data-mining technologies can now provide police with vast amounts of surveillance information and could radically increase police power. Policing can be increasingly targeted at specific people and neighborhoods — with potentially serious inequitable effects.
One speaker at a recent national law enforcement conference compared future police work to Minority Report, the Tom Cruise film set in 2054 Washington, where a “PreCrime” unit has been set up to stop murders before they happen.
While PreCrime remains science-fiction, many technology advances are already involved with predictive policing — identifying risks and threats with the help of online information, powerful computers and Big Data.
New World Systems, for example, now offers software that allows dispatchers to enter in a person’s name to see if they’ve had contact with the police before.  Provided crime data, PredPol claims on its website that  its software “forecasts highest risk times and places for future crimes.” These and other technologies are supplanting and enhancing traditional police work.
Public safety organizations, using federal funding, are set to begin building a $7-billion nationwide first-responder wireless network, called FirstNet. Money is now being set aside. With this network, information-sharing capabilities and federal-state coordination will likely grow substantially. Some uses of FirstNet will improve traditional services like 911 dispatches. Other law enforcement uses aren’t as pedestrian, however.
One such application is Beware, sold to police departments since 2012 by a private company, Intrado. This mobile application crawls over billions of records in commercial and public databases for law enforcement needs. The application “mines criminal records, Internet chatter and other data to churn out … profiles in real time,” according to one article in an Illinois newspaper.
Here’s how the company describes it on their website:
Accessed through any browser (fixed or mobile) on any Internet-enabled device including tablets, smartphones, laptop and desktop computers, Beware® from Intrado searches, sorts and scores billions of commercial records in a matter of seconds-alerting responders to potentially deadly and dangerous situations while en route to, or at the location of a call.
Crunching all the database information in a matter of seconds, the Beware algorithm then assigns a score and “threat rating” to a person — green, yellow or red. It sends that rating to a requesting officer.
For example, working off a home address, Beware can send an officer basic information about who lives there, their cell phone numbers, whether they have past convictions and the cars registered to the address. Police have had access to this information before, but Beware makes it available immediately.
Yet it does far more — scanning the residents’ online comments, social media and recent purchases for warning signs. Commercial, criminal and social media information, including, as Intrado vice president Steve Reed said in an interview with urgentcomm.com, “any comments that could be construed as offensive,” all contribute to the threat score.
There are many troubling aspects to these programs. There are, of course, obvious risks in outsourcing traditional police work — determining who is a threat — to a proprietary algorithm. Deeming someone a public threat is a serious designation, and applications like Beware may encourage shortcuts and snap decisions.
It is also disconcerting that police would access and evaluate someone’s online presence. What types of comments online will increase a threat score? Will race be apparent?
These questions are impossible to answer because Intrado merely provides the tool — leaving individual police departments to craft specific standards for what information is available and relevant in a threat score. Local departments can fine-tune their own data collection, but then threat thresholds could vary by locale, making oversight nearly impossible.
Tradition holds that justice should be blind, to promote fairness in treatment and avoid prejudgment. With such algorithms, however, police can have significant background information about nearly everyone they pull over or visit at home. Police are time-constrained, and vulnerable populations – such as minorities living in troubled neighborhoods and the poor — may receive more scrutiny.
No one wants the police to remain behind a thick veil of ignorance, but invasive tools like Beware — if left unchecked — may amplify the current unfairness in the system, including racial disparities in arrests and selective enforcement.
Intrado representatives defend Beware’s perceived intrusiveness, pointing out that credit agencies have similar types of information. This data-mining program, however, goes beyond financial records to include social media, purchases and online comments when assigning a rating.
And no system is foolproof. Congress, for example, recognizes the sensitivity of the information that lenders and employers have, because errors can cause serious financial harm. The Fair Credit Reporting Act therefore gives consumers the right to access their credit reports and make corrections.
The risks to life and property, however, are far higher and more unpredictable in the law enforcement context. Yet there is no mechanism for people to see their threat “ratings” — much less why the algorithm scored it. You have no ability to correct errors if, say, someone with the same name has a violent criminal record.
Another effect is that these technologies give law enforcement the ability to routinely monitor obedience to regulatory minutiae and lawmaker whims. Police officers now boast, for example, that the Beware system allows the routine code enforcement of a nanny state — such as identifying homeowners so overgrown trees on a property can be trimmed.
Beware can also encourage fishing expeditions and indiscriminate surveillance in the hopes of finding offenders. Police used Beware recently at a Phish concert in Colorado, for example, checking up on concertgoers based on car license plates.
Perhaps the most serious issue is that such systems may be used as pretext in unconstitutional investigations. John Shiffman and Kristina Cooke reported for Reuters last year that a secretive Drug Enforcement Administration unit regularly funnels information to other law enforcement agencies in order to launch criminal investigations. This information is frequently acquired via intelligence intercepts, wiretaps and informants. As the FirstNet national wireless network rolls out, federal-state coordination will likely increase opportunities for police to receive sensitive information from powerful federal agencies.
Data-mining gives police significantly more information to create reasonable suspicion for suspects that federal agencies flag. Officers could receive a search or arrest warrant with the help of information gleaned from Beware and other databases, like those tracking license plates. If an arrest follows, data-mining helps provide the police with the legal pretext to engage in these fishing expeditions. Defendants will likely have no opportunity to challenge the legality of the original surveillance that led to their arrest.
As predictive policing investment ramps up, and local police and federal agencies increasingly coordinate, more secrecy becomes more valuable. Local police and prosecutors often refuse to disclose how they gain information about defendants because federal agencies prohibit them from discussing these technologies. In Baltimore, for example, police recently dropped evidence against a defendant rather than reveal information about cellphone tracking that the FBI did not want disclosed in court.
Yet police might not acquire some of this equipment if the local community is made fully aware of its use. Consider, the city council of Bellingham, Wash., recently rejected a proposed purchase of Beware. The police department had applied for, and received, a one-time $25,000 federal grant to cover some of the $36,000 annual cost of Beware. At a mandatory hearing about the purchase, Bellingham citizens discovered how Beware worked and opposed the purchase because of both the cost and the privacy implications. The funds were subsequently redirected.
This rejection demonstrates that many modern policing techniques — and the accompanying secrecy — can antagonize the average citizen. The occasional appearance of sniper rifles and military vehicles only stokes that sentiment. Local police forces increasingly receive military surplus equipment and federal lucre from an alphabet soup of U.S. agencies and opportunistic contractors. Now police are using, typically without residents’ knowledge, powerful databases, along with cellphone and license-plate trackers.
Police need guidance about under which circumstances these sophisticated databases can be used. An inaccurate threat level for a residence, after all, can change how police approach a situation. Failure to update who lives at a particular residence, for example, could transform a green rating into a red rating — turning a midday knock on the front door into a nighttime SWAT raid.

PHOTO (TOP): Tom Cruise in Minority Report. Courtesy of  20th Century Fox
PHOTO (INSERT ): The dashboard for the New York Police Department’s ‘Domain Awareness System’ is seen in New York, May 29, 2013. REUTERS/Shannon Stapleton

PHOTO (INSERT): Police officers point their weapons at demonstrators protesting against the shooting death of Michael Brown in Ferguson, Missouri, August 18, 2014. REUTERS/Joshua Lott

No comments:

Post a Comment