DHS biometrics collected from migrant kids could solve FRT training dataset shortfalls

An investigation by MIT Technology Review has revealed a plan by the U.S. Department of Homeland Security to collect and analyze photos of the faces of migrant children at the border, in order to train and improve facial recognition technology.

The government says the idea is to improve biometric identity services for migrants. Speaking at a conference in June, John Boyd, assistant director of DHS’ Office of Biometric Identity Management (OBIM), framed the key question as, “If we pick up someone from Panama at the southern border at age four, say, and then pick them up at age six, are we going to recognize them?”

The answer is, yes – if you collect craniofacial structural progression data from the faces of thousands of migrant children. The Review points to numbers from Syracuse University’s Transactional Records Access Clearinghouse (TRAC), showing that 339,234 children arrived at the U.S.-Mexico border in 2022. Around 150,000 were unaccompanied. “If the face prints of even one percent of those children had been enrolled in OBIM’s craniofacial structural progression program, the resulting data set would dwarf nearly all existing data sets of real children’s faces used for aging research.”

In response to MIT’s revelation, the DHS sent an email specifying that it “uses various forms of technology to execute its mission, including some biometric capabilities,” ensuring “all technologies, regardless of type, are operated under the established authorities and within the scope of the law.”

Consent impossible for kids in high-pressure context, say rights groups

If not everyone is surprised at the news – the Review quotes a former Customs and Border Patrol official who says every migrant processing center he ever visited was conducting wholesale biometric identity collection, children included – there is inevitable pushback around issues of privacy and consent.

In comments to the Review, a representative for Senator Jeff Merkley of Oregon says “the use of facial recognition technology to track migrant children is another stride toward a surveillance state.” Civil rights and child safety advocates worry that the “vast power differentials” between U.S. border authorities and unaccompanied migrant children complicates the issue of informed consent.

Ashley Gorski, a senior staff attorney with the American Civil Liberties Union (ACLU), tells the Review that DHS “should have to meet an extremely high bar to show that these kids and their legal guardians have meaningfully consented to serve as test subjects,” given the “significant intimidation factor.”

DHS on decades-long biometric data collection spree

Like DHS itself, OBIM was created in the wake of the 2001 terrorist attacks on the World Trade Center. Over time, it has increased its mandate and activity, collecting face prints, iris and retina scans, and DNA.

As illegal border crossings have crystallized into a sharp, divisive political issue, border agencies have ramped up biometric data collection programs, exploring new contactless methods and releasing the CBP One, which allows asylum seekers to submit face biometrics in advance of crossing. The DHS now collects biometrics and DNA from migrants as soon as they cross the border, resulting in the addition of more than a million DNA profiles into law enforcement databases since collection began in January 2020.

CBP One has faced allegations of discriminatory programming (notably facial recognition algorithms that are unable to recognize darker-skinned people). Should errors or structural bias result in an overrepresentation of people of color in DHS databases, they could face unwarranted discrimination or over-policing.)

To date, DHS has mainly limited its biometric collection to those aged 14 to 79. But it has pushed at that limit before, and would evidently like to be able to collect biometrics from minors, given its penchant for collecting biometrics in general. Finally, the Review speculates that because OBIM’s scheme is conducted under the auspices of the DHS’s undersecretary of science and technology, it is categorized as being “for research,” rather than to be used in DHS operations – meaning many of the standard restrictions for DHS use of facial recognition and face capture technologies do not apply.

Given the opportunity by the MIT Technology review to comment, Boyd says what the agency is trying to do is fill gaps in database training data by getting “large data sets of known individuals” taken “under controlled conditions.”

Synthetic data better for privacy but still has flaws, says Schuckers

De-aging adult faces, or generating synthetic face data to represent different age cohorts, has been proposed as an alternative to mass biometric data collection. But, while the technology is advancing, it’s still not going to provide the same results as real face data.

Shuckers’ Center for Identification Technology Research (CITeR) at Clarkson University has been researching a more patient method, collecting face biometrics from 231 volunteers in elementary and middle schools every six months since 2016. But the kinds of conditions required for this longitudinal study are impractical, or impossible, at the border.

Border is ‘perfect laboratory for tech experimentation’: Petra Molnar

Petra Molnar, author of The Walls Have Eyes: Surviving Migration in the Age of AI, tells the Review that borders are “the perfect laboratory for tech experimentation, because oversight is weak, discretion is baked into the decisions that get made.” Her point is not just that a massive power imbalance is at play in collecting biometric information from migrants – but also that, once a technology is established, it tends to creep outward from its initial use case.

MIT backs her up in noting that John Boyd had a large hand in setting up biometric operations in Iraq, which were used to identify and target insurgents – and which “subsequently played a substantial role in influencing the expansion of biometric data collection by the Department of Defense, which now happens globally.”

Article Topics

biometrics  |  border security  |  children  |  data collection  |  dataset  |  DHS  |  face biometrics  |  facial recognition  |  OBIM  |  training