Site icon Sophos News

Tattoo recognition technology “raises significant First Amendment questions”

An investigation by the Electronic Frontier Foundation (EFF) is shining a spotlight on a US government program developing algorithms that can identify similar tattoos from images, for use by the FBI and law enforcement.

The National Institute of Standards and Technology (NIST) has been conducting research into tattoo recognition technology since 2014, relying on a database of 15,000 tattoo images collected by the FBI from prisoners and arrestees without their consent, according to the EFF.

Tattoos, which are usually elective (people choose their own tattoos), can reveal a person’s cultural, religious and political beliefs, the EFF says, raising concerns about how this technology impacts First Amendment rights to freedom of speech and religion.

The algorithms developed by the NIST could be used by law enforcement to link individuals in gangs, but also people affiliated with the same religious or political groups, the EFF said in a report released last week.

Eventually, the EFF says, the algorithms could be used to track individuals by matching images in a database with images of tattoos captured by police or surveillance cameras.

Many of the tattoos in the FBI database showed Christian or Catholic iconography, such as crosses, rosary beads or images of the crucifixion, which the EFF said was “totally inappropriate.”

An NIST slide obtained by the EFF through a public records request showed the NIST touted tattoo recognition as a way to link individuals to “gangs, sub-cultures, religious or ritualistic beliefs, or political ideology.”

The EFF also raised concerns about the privacy of individuals whose tattoos were included in the database used in the NIST’s research – not just inmates, but people who were arrested or merely stopped by police.

The tattoo images were shared with private companies such as MorphoTrak, “one of the largest marketers of biometric technology to law enforcement agencies,” including personally identifiable information such as names, faces and birth dates, the EFF report says.

Furthermore, the NIST research is being conducted without scrutiny or ethical oversight, the EFF claims:

Under federal research guidelines, research involving prisoners triggers enhanced scrutiny and ethical oversight to prevent their exploitation. Instead, NIST and the FBI are treating inmates as an endless supply of free data.

In a written response to the EFF report, the NIST said its research will “help ensure tattoo matching technologies are evaluated using sound science to improve accuracy and minimize mismatches.”

Despite the concerns raised by the EFF report about ethical research practices and data sharing, the NIST denied that the FBI tattoo database contained any other identifying information, and said its research does not come under federal regulations for human-subject research.

Another round of NIST research will seek to gather another 100,000 tattoo images collected by police and prison agencies in Florida, Michigan and Tennessee.

The EFF said the government should “act responsibly and halt the program until it goes through the full oversight process.”

Tattoos can already be used by law enforcement to identify individuals, of course, but the use of image recognition technology raises civil liberties concerns about how the images are gathered, used and stored.

Law enforcement agencies are increasingly tracking people by biometrics, whether it’s facial recognition, fingerprints, voiceprints or genetic material including DNA obtained from blood or a person’s unique cloud of microbes.

The FBI and many state and local police departments in the US have for years been creating their own databases of criminal suspects’ DNA, without any kind of regulation or transparency as to how the information should be used or for how long.

The EFF and MuckRock have undertaken a national census to gather public records on what biometric technologies state and local law enforcement agencies have and how they’re using them.


Exit mobile version