TECHNOLOGY’S GREAT – UNTIL IT’S NOT
Police love Rapid DNA and facial recognition but hate encryption.
Privacy advocates beg to differ.
For Police Issues by Julius (Jay) Wachtel. DNA’s forensic applications date back to 1984, when Dr. Alec Jeffries discovered that he could distinguish between family members by comparing repeat sequences of linked chemical pairs in their genetic code. Within a couple of years he used the technique to assist police, and in the process helped free an unjustly accused man. DNA’s usefulness for fighting crime (and exonerating the innocent) quickly became evident, and its use took off.
Forensic DNA analysis is couched in probabilities. An absolutely positive match, with the likelihood of error less than one over the population of the Earth, requires that samples have identical linked pairs at each of thirteen standard locations (“loci”) in the human genome. (For more on this see our interview with a real expert at “DNA: Proceed With Caution.”) Attaining that level of certainty is not difficult when DNA is abundant, say, from a cheek swab. But things can get tricky with crime scene evidence, and especially when the DNA is a mix with multiple contributors.
Processing DNA involves four steps: extracting whatever DNA may be present; measuring its quantity and assessing its quality; using the “PCR” method to make millions of copies of DNA sequences; and finally, separating the DNA molecules, reducing their characteristics to a profile that can be compared with other samples. Each step involves different groups of highly-skilled technicians laboring in clean rooms using specialized tools and expensive machines. Even under the most favorable conditions, and with the most up-t0-date equipment, traditional DNA typing can consume an entire day.
Click here for the complete collection of technology & forensics essays
That’s why the notorious backlog. Despite an infusion of local, state and Federal funding, the reported “average” wait for results is – hold on! – five months. (For information about the Fed’s DNA “backlog reduction program” click here.)
Now imagine a lone operator doing it all in less than two hours with a single machine! That’s what “Rapid DNA” is all about. First used in criminal casework by the Palm Bay, Florida police department, Rapid DNA has been adopted by many local and state agencies, most recently the Kentucky State Police. So far, though, the newfangled machines work best with substantial quantities of single-source DNA (i.e., cheek swabs) and tend to be flummoxed by small or mixed samples, such as “unknown” DNA found at crime scenes. Accordingly, Rapid DNA profiles are not considered sufficiently trustworthy for inclusion in or comparison with the approximately eighteen million DNA profiles of local, state and Federal arrestees and convicted persons present in the FBI’s national CODIS database, which were processed in the old-fashioned, time-consuming way.
However, a major Federal initiative is underway to upgrade Rapid DNA technology so that its output is acceptable for CODIS. If it succeeds, and DNA typing becomes, as the FBI’s “infographic” tantalizingly suggests, a routine aspect of the booking process, police could quickly compare crime scene DNA with genetic profiles of most anyone who’s run seriously afoul of the law. And yes, its use to that end has been endorsed by the “Supremes.” Here’s what our nation’s highest court said in Maryland v. King (2013):
When officers make an arrest supported by probable cause to hold for a serious offense and bring the suspect to the station to be detained in custody, taking and analyzing a cheek swab of the arrestee’s DNA is, like fingerprinting and photographing, a legitimate police booking procedure that is reasonable under the Fourth Amendment.
At present, U.S. Government agencies collect DNA “from individuals who are arrested, facing charges, or convicted, and from non-United States persons who are detained under the authority of the United States, subject to certain limitations and exceptions” and “to preserve biological evidence in federal criminal cases in which defendants are under sentences of imprisonment.” Each State also permits or requires collecting DNA from persons convicted of felonies, and thirty authorize it for certain classes of arrested persons. California, for example, allows DNA typing of everyone arrested for a felony and of a misdemeanor should they have a prior felony conviction.
Although there have been problems with DNA technology, and the legal authority to collect it is occasionally challenged, even the staunchest civil libertarians will concede that when it comes to wrongful arrest and conviction DNA has been an overwhelming force for good. But the advent of Rapid DNA – read, “cheap and quick” – has enabled the technique’s use in politically controversial areas.
Such as immigration enforcement. DHS, for example, recently deployed Rapid DNA at the Southern border to combat fraudulent claims by would-be immigrants that the children who accompany them are blood relatives. Its successful use to confirm parentage inspired a Federal initiative that would collect DNA from everyone in immigration custody and transmit it to CODIS. That goal, which was clearly unattainable until Rapid DNA came on scene, was first authorized by a 2008 Federal rule that sought to use DNA “to solve and hold [unauthorized immigrants] accountable for any crimes committed in the United States…before the individual’s removal from the United States places him or her beyond the ready reach of the United States justice system.” But the ACLU balked. Here’s what one of its lawyers said:
That kind of mass collection alters the purpose of DNA collection from one of criminal investigation basically to population surveillance, which is basically contrary to our basic notions of a free, trusting, autonomous society.
Similar concerns have been voiced about the emerging field of facial recognition. Unlike DNA, the technology used to identify persons from photographs and video images isn’t sufficiently trustworthy to use in court. Instead, it’s all about generating investigative leads. Thanks to artificial intelligence software, detectives with, say, a robbery video can quickly scour photo databanks – think passports, driver licenses, criminal record repositories, online social networks and so on – for a match. According to a recent NBC news investigation, that approach can yield impressive results:
In Colorado, local investigators have foiled credit-card fraudsters, power-tool bandits and home-garage burglars and identified suspects in a shooting and a road-rage incident…The technology has led to the capture of a serial robber in Indiana, a rapist in Pennsylvania, a car thief in Maine, robbery suspects in South Carolina, a sock thief in New York City and shoplifters in Washington County, Oregon.
It’s not just about conventional crime. Driver licenses are the predominant tool of everyday identification, so it’s important to prevent their fraudulent issue. Motor vehicle agencies (New York State is one) have used facial-recognition technology for years to combat fraudsters who amass multiple licenses under different names.
Yet doubts about facial recognition linger. Accuracy-wise, it’s got a ways to go. Even after all the development, it still works best for white males. Erroneous matches for persons of color and, especially, black women, are depressingly common. But just like Rapid DNA, the sharpest criticism is about the threat that facial recognition poses to civil liberties. “The Perpetual Line-Up,” a project of Georgetown Law School, warns that there is “a real risk that police face recognition will be used to stifle free speech”:
Of the 52 agencies that we found to use (or have used) face recognition, we found only one, the Ohio Bureau of Criminal Investigation, whose face recognition use policy expressly prohibits its officers from using face recognition to track individuals engaging in political, religious, or other protected free speech.
Alarm that America will mimic China, where video surveillance is ubiquitous, have led the liberal burg’s of Berkeley, Oakland, San Francisco and Sorverville, Massachusetts to ban government use of facial recognition technology altogether. Even Detroit has imposed strict controls. A crime-fraught place where more than four-hundred businesses beam continuous video to police, facial recognition is limited to still photos and violent crimes. So forget about using those video streams! Really, had it not been for Police Chief James Craig’s supplications to the city council – he promised that police would compare photos with exquisite care – America’s one-time “Motor City” would have likely imposed a total, Berkeley-like ban. Still, his mention that facial recognition had already identified many violent criminals enraged activists, who were upset that the technology had been employed without their assent.
Most recently, facial recognition has taken a hit over its use in – you guessed it – immigration enforcement. According to the Washington Post, more than a dozen States are issuing driver licenses or their equivalents to illegal immigrants. Immigration authorities know that, so they regularly turn to those states’ DMV databases to run the photos of, say, absconders who fail to show up for their hearings. That’s drawn Georgetown Law’s ire. According to Clare Garvier, who led the school’s research, “it’s an insane breach of trust” for DMV offices to knowingly issue licenses “then turn around and allow ICE access” to the photos.
Technology has given police new tools. It’s also taken some away. In “A Dead Man’s Tales” we wrote of the FBI’s struggle to get Apple to give up the passcode for a dead terrorist’s cell phone. (Eventually, the Feebs managed to unlock the device on their own.)
Be sure to check out our homepage and sign up for our newsletter
But law enforcement’s concerns aren’t only about “national security.” The Internet is a known go-to place for a multitude of odious pursuits. Say, child sex trafficking. Yet service providers seem strangely indifferent. As Exhibit A, consider Facebook’s plan to institute end to end encryption for all its messaging services (WhatsApp, Instagram and Facebook Messenger). Determined to make both sides to conversations impermeable to real-time interception (what we used to call “wiretapping”), humanity’s most popular social media platform stubbornly resists the notion of incorporating an electronic “back door.”
How to move forward? A recent Carnegie report about the struggle suggests that cops forego intercepting “data-in-motion,” that is, as it flows in real time. In exchange, they could get (with a warrant, of course) a pass-key for accessing “data-at-rest”, meaning what’s present on devices they seize. Of course, from the law enforcement perspective that’s hardly sufficient. Stripped of the ability to act proactively, police and the Feds would have to content themselves with collecting evidence after the fact; that is, what the bad guys and girls didn’t erase. Preventing crime? Minimizing potentially horrific outcomes? Forgedabboud it!
And that “solution,” dear readers, is what preeminent members of the American intelligentsia propose. Good thing they’re not crooks!
7/1/21 A new Maine law prohibits the use of facial recognition technology except when there is probable cause to believe than an imaged person “committed a serious crime.” Police may then request a comparison to FBI and motor vehicle databases. Matches do not by themselves permit making an arrest. Requests “will be tracked” and individuals who feel their rights were violated may sue. Meanwhile Southern California police departments are being criticized for posting images of wanted persons on Facebook and Instagram. While these have led to arrests, many posts are for minor crimes and remain online even after suspects have been arrested or cleared, damaging their privacy and future prospects.
6/30/21 Mounted on poles and on police cars, license plate cameras record the license plate number of passing vehicles and instantly compare them to national, state and local lists of vehicles that are stolen, were reportedly involved in a crime, or are being driven by wanted persons. That’s led to many arrests. It’s also led to concerns that information collected by these systems can be resold or misused.
6/16/21 In a major policy address, A.G. Merrick Garland identified “racially or ethnically motivated violent extremists, and militia violent extremists” as America’s major domestic terrorism threats. Their access to social media, encrypted communications and firearms poses significant challenges. But DOJ’s enforcement focus will remain on violence, not ideology. “In America, espousing a hateful ideology is not unlawful. We do not investigate individuals for their First Amendment-protected activities.”
6/4/21 How best to check the identities of the “tens of thousands” of asylum-seekers awaiting entry to the U.S.? As part of the solution, DHS has implemented a facial-recognition app. that allows CBP to compare images of would-be entrants to persons already in its database and confirm whether they have officially applied or are part of the subset who may enter despite the COVID-19 policy that’s kept them out. But civil libertarians worry about the app’s security and its potential for error and misuse.
6/3/21 After its use to identify and arrest a protester, the D.C. Council of Governments halted the use of NCRFRILS, a facial-recognition system used by D.C.-area law enforcement agencies. Police argued that the system had helped solve robberies and other crimes, but civil libertarians cited its potential to wrongly identify women and minorities. In Virginia, police use of Clearview AI spurred a new State law that requires pre-approval before any facial recognition system can be used in the future.
5/19/21 Fearing that use of its facial recognition software, Rekognition, had a potentially chilling effect on public discourse, unfairly affected minorities, and could lead to wrongful convictions, Amazon suspended its use by law enforcement agencies last year. Amazon said it would await the passage of legislation to regulate the use of this tool. It’s now extended the prohibition indefinitely.
3/1/21 Fears of bias against non-Whites has led to bans against police use of facial recognition technology in cities including Oakland, Portland, San Francisco and Minneapolis. However, its many successful outcomes led Massachusetts to include a provision in a police reform bill that requires police to seek court orders to perform facial recognition searches except in emergencies.
2/27/21 Privacy advocates are criticizing ICE for subscribing to the CLEAR commercial database which compiles “millions” of telephone, electricity and other utility subscriber records. ICE investigators use this information to help identify and track down immigration law violators.
12/8/20 Authorities identified “Golden State Killer” Joseph De Angelo by uploading his profile to the MyHeritage consumer DNA database, which produced a match to a close relative. But MyHeritage, whose “terms of service” cautioned that it responded to “official inquiries,” wasn’t informed that police were behind the search.
11/18/20 LAPD has officially prohibited the use of any facial recognition database for generating investigative leads other than Los Angeles County’s “Regional Identification System” system. This move follows on a Buzzfeed news article that reported LAPD detectives were frequently turning to a commercial system, Clearview AI, which “scrapes” images from the Internet, including social media. Civil liberties advocates have condemned its use because of concerns about privacy and accuracy.
10/13/20 DOJ issued a stern warning about civilian use of drones, calling them “an amazing technology that offer great commercial promise, but they also present a serious challenge to ensuring public safety.” Drone operators have been prosecuted for, among other things, intruding into restricted airspace, flying over civil disturbances, dropping drugs into a prison, and dropping explosives as an act of terrorism.
In 2019 the FBI received over 450,000 reports of cybercrime, with losses totaling over $3.5 billion. Among the most reported scams were phishing (ruse that gets personal info.), non-payment/non-delivery, extortion, personal data theft, business e-mail compromise, confidence fraud romance, and identity theft. Click here for the FBI 2019 Internet Crime Report.
10/12/20 Officials in the U.S., Great Britain, Australia, India and Japan issued a joint statement calling for regulations to allow governments to access end-to-end encrypted communications when there is evidence of “child sexual exploitation and abuse, violent crime, terrorist propaganda and attack planning.” Technology providers are urged to build in means of access and to consider user safety when designing their products.
9/22/20 Los Angeles police regularly turn to a regional facial recognition database maintained by the county sheriff’s department to develop leads on suspects of crime. In a recent success officers arrested a man suspected of sexually assaulting children. Although police say they are aware of the software’s limitations, the ACLU disparaged its use last year with a test that identified one in five lawmakers as a criminal. Its purpose was to support a proposed State law that would ban its use on body cameras.
9/18/20 Remarks delivered by DOJ to State attorney generals emphasize that in today’s online world it is vital that law enforcement agencies be given lawful access to encrypted communications. As one example the agency offered the crisis of child exploitation, which is facilitated by the ready availability of locked and encrypted cell phones that prevent access even with a search warrant.
9/16/20 An elaborate statement issued by the Department of Justice emphasized the usefulness of facial recognition technology in the fight against crime. DOJ pledged to use the technology, and assure that partner agencies do so, in a manner that “minimizes inaccuracy and unfair biases” and respects the interests of liberty and privacy.
8/13/20 Clearview A.I. has assembled a database of “billions” of facial images from the Internet, including from popular services such as Instagram. It uses these to compare against facial images furnished by police to help them identify and track down persons suspected of crime. Several state attorney generals are suing Clearview for violating privacy laws. But the company claims that its activities are protected under the First Amendment.
7/3/20 Police in California and four other states are involved in an FBI test of Rapid DNA for booking arrestees. In Orange County, the D.A.’s office has used it for five years, building a database of profiles from misdemeanants as part of their plea bargains. But Rapid DNA profiles remain insufficiently trustworthy for inclusion into CODIS. Mixed samples, in particular, still produce inconsistent results.
6/26/20 Santa Cruz, Calif., an early adapter of Predictive Policing, has banned it because it biases police attention towards areas populated by persons of color. Its use was suspended by a new police chief in 2017 because doing “purely enforcement” caused inevitable problems with the community. Santa Cruz also banned facial recognition software because of its racially-biased inaccuracies.
6/25/20 Detroit police arrested a black man in a store theft. Software tentatively matched a still image taken from a surveillance video to his photo in the state’s driver license database. Detectives included this image in a six-pack and showed it to the store’s security contractor, who wasn’t present during the theft. She identified him, and police used that alone to obtain a warrant. But the man was innocent.
6/24/20 A new Senate bill, “The Lawful Access to Encrypted Data Act,” would require that manufacturers and providers of communication services help law enforcement agencies access encrypted devices and decode encrypted messages if ordered by a court based on probable cause.
4/11/20 Objections by civil rights activists led a Baltimore judge to put the brakes on a test program to have small planes shoot aerial footage of the city for 40 hours per week. According to police, the results would only be used to help solve violent crimes, of which the city has in abundance.
3/6/20 DOJ announced that ICE will soon begin routinely collecting DNA from “certain” persons who are detained for illegal entry. As is the case for ordinary arrestees, these samples will be included in the FBI’s national CODIS database. Officials are preparing a rule that will identify exactly whom among the scores of illegal migrants the new procedure will affect.
2/10/20 Companies aggressively marketed facial recognition, artificial intelligence and other technologies to police at the annual European Police Congress, held in Berlin..
2/10/20 Drawn from social media and other publicly-available sources, Clearview’s vast image database is helping participating police agencies identify victims of child abuse depicted on offender videos and other media. While the approach has met with success, privacy advocates fear its misuse.
1/25/20 London police announced that video cameras using real-time facial recognition technology will be deployed around the city to help locate suspects of serious crimes as they walk the streets. Meanwhile, concerns about privacy have led New Jersey’s attorney general to bar local prosecutors and police from using Clearview.
1/24/20 In a letter accusing the A.I. provider of violating Twitter’s terms of service, the social media website demanded that Clearview stop using anything on its platform and purge its files (see 1/18/20 update)
1/18/20 Hundreds of U.S. law enforcement agencies have subscribed to Clearview, a year-old service provider that uses proprietary A.I. technology to compare suspect photos against a rapidly-growing database populated with millions of images drawn from websites including Facebook and YouTube. Its first successful application for police, in February, took twenty minutes to match the face of an unidentified shooter, captured on a bystander’s cell-phone video, to his image on social media.
1/8/20 Saudi military student Mohammed Alshamrania, who shot and killed three and wounded eight at a Florida naval base December 6, had two Apple iPhones. The FBI secured warrants to search the phones but found that both are password-protected. So it’s reached out to Apple. But Apple’s long insisted it can’t access locked phones, and has so far refused to build in a “back door.”
12/12/19 Democratic and Republican lawmakers joined with Attorney General William Barr and British and Australian officials to demand that Internet portals build in “backdoors” that allow law enforcement agencies to intercept encrypted communications. But Facebook, which is planning end-to-end encryption, complained that doing so would make its system vulnerable and be “a gift to criminals, hackers and repressive regimes.”
11/15/19 A proposed Federal law would require Federal agents to get a warrant to use facial technology recognition to track a suspect for more than three days. It would not limit its typical current use, which is to make matches after-the-fact. Civil libertarians are demanding that this latter practice, which is now commonplace, face similar oversight.
Did you enjoy this post? Be sure to explore the homepage and topical index!
Home Top Permalink Print/Save Feedback
Preventing Mass Murder Is Your Uncle a Serial Killer? A Dead Man’s Tales
State of the Art – Not! Labs Under the Gun