In recent years, police have increasingly used data from ancestry-testing companies to identify major crime suspects through the DNA of their family members. A new case, however, is reigniting the debate about when and where it’s appropriate to use this powerful crime-fighting tool.
Famously, genetic genealogy was credited with identifying the man suspected of being the Golden State Killer and serial rapist last year, making a compelling argument for the forensic use of familial DNA. In that case, investigators uploaded crime scene DNA to GEDMatch, an open source database made up of test results shared by customers of services like 23andMe and Ancestry.com. Since then, a number of law enforcement agencies have turned to data from ancestry-testing companies to investigate serious unsolved crimes.
Recently, however, familial DNA was used to nab a teen believed to have attacked a 71-year-old woman at a chapel in Utah, which Buzzfeed News reports is the first time this tech has been used to identify a violent assault suspect.
The woman was reportedly playing organ inside a locked Mormon church last November when she heard “several minutes” of loud pounding on the chapel’s door, which she did not open. According to a recently unsealed search warrant obtained by the Deseret News, the assailant then broke a window and entered the chapel before pulling the woman off her organ bench and strangling her “to the point that she lost consciousness,” possibly multiple times.
The suspect reportedly left behind blood at the scene, and detectives sent those DNA samples to a crime lab to be tested. When they received no hits against the federal database, a detective then began exploring genealogy in hopes of tracking down a possible lead. Detectives sent their DNA samples to Parabon NanoLabs, which confirmed to Gizmodo that it performed genetic genealogy for the case.
But that’s where things get tricky. According to BuzzFeed News, Parabon initially told detectives that it could not use the DNA research hub GEDmatch—the website used by police to identify the suspected Golden State Killer—because of the specific rules around which law enforcement is allowed to use the free tool. According to its terms of service, users agree that by uploading their DNA profiles to GEDmatch, that information may be obtained by law enforcement to identify perpetrators of violent crimes, which are limited to homicide and sexual assault. The strangling at the chapel was neither of these.
BuzzFeed News reports that a detective with the case approached GEDmatch co-founder Curtis Rogers for express permission, which he granted. Using familial DNA, police were able to narrow their search down to a 17-year-old who lived near the chapel where the attack occurred. Police were able to retrieve a juice box and a plastic milk container that the teen had tossed after lunch and used them to match the teen’s DNA to that left at the crime scene. The teenager, who has not been identified, was arrested in April and booked into a youth detention center.
There are a number of issues at play with this case, the most immediate being the implications for users who understood their genetic profiles could be used by law enforcement under specific circumstances—which now appear more muddied.
Speaking with the Associated Press last year about the Golden State Killer case, Rogers said the investigation was “done without our knowledge, and it’s been overwhelming.” He also told the news service that GEDmatch doesn’t “hand out data,” though that appears to be what happened in this most recent case. Moreover, Rogers’ own parameters for what qualifies under the service’s “violent crimes” rule seems murky and susceptible to interpretation, which creates a precarious privacy situation, even if users are knowingly uploading their genetic profiles under the assumption that they may be used to aid police under specific circumstances.
Speaking with BuzzFeed News, Rogers said he permitted the use of his service for “this one case” on the grounds that it was “as close to a homicide as you can get.” A spokesperson for GEDmatch did not immediately return a Gizmodo request for comment about its use in the Utah investigation.
When asked about the GEDmatch’s use in the case and how that was communicated to its users, Parabon CEO Steve Armentrout told Gizmodo by phone that both companies acknowledged that it was an exception to the site’s terms of service.
“This was the first and only time we’ve done this and Curtis and the GEDmatch folks put something up on the website acknowledging what happened,” he said.
As with the Golden State Killer and Utah cases, Parabon was also used to help identify the alleged killer of a newborn baby who was left in a ditch in and died of exposure Sioux Falls, South Dakota in 1981. Police identified a married woman named Theresa Rose Bentaas, a mother of two adult children, as the infant’s killer. Police revealed DNA appears to indicate she was also the child’s mother.
According to Inside Edition, Bentaas reportedly became pregnant with the child at a very young age by the man who would later become her husband, carried the pregnancy to term in secret, and delivered the baby alone before abandoning the infant. Parabon provided the genetic profile that eventually led police to Bentaas, who has been charged with first-degree murder, second-degree murder, and second-degree manslaughter.
Detective Michael Webb, who worked the case, told reporters that no one is using Parabon “for small cases… these are cold cases, these are unidentified human remains that they’ve never been able to identify.” But some argue the Bentaas case blurs the line between clear use cases for forensic genetic genealogy and something that’s more difficult to define.
“A Baby Doe case is a special category,” Colleen Fitzpatrick, co-founder of nonprofit DNA Doe Project, told Wired in March. “It breaks apart this neat dichotomy between finding criminals who might still be dangerous and finding victims.”
Speaking to the gray area introduced by the case in Utah, Armentrout told Gizmodo that his personal feeling is that “a more nuanced definition of violent crime would be both useful and palatable to most people on GEDmatch and most American citizens.” He pointed a PLOS One survey conducted last year that found 91 percent of respondents felt law enforcement should be allowed to use genealogical databases to help solve violent crimes, defined by the authors as including rape, murder, arson, or kidnapping. (That figure fell sharply to 46 percent for nonviolent crimes, such as drug possession.)
“Think about that. It’s hard to get 91 percent of Americans to agree on anything,” Armentrout said. “It seems to me that the GEDmatch definition of violent crime is sometimes restrictive. We’ve declined to do genetic genealogy on cases that are clearly heinously violent based on that strict definition and the way it’s defined now. I would be in favor of a conversation among the community about a more nuanced definition of violent crime because the intent to commit to murder—failed attempts where people are maimed for life—clearly needs to be investigated with all the tools in the arsenal. But the terms of service right now don’t allow that.”
Whether or not users agree, these cases raise serious questions about who gets to decide when police are able to use ancestry-testing data to investigate crimes. In Utah, it seems, the genetic privacy of a million GEDmatch users (and an untold number of family members) came down to one man’s decision. And if private individuals are making these choices, where will they choose to draw the line? It’s likely these cases are just scratching the surface of how forensic genetic genealogy will change our lives, and that alone should alarm you.