Police collect and use all kinds of data to describe suspects, like eye and hair color, height, body type, race and other physical details. Tattoos can also be useful to law enforcement agencies because they are often distinct, and they often have meaning.
The National Institute of Standards and Technology (NIST) is a government agency that conducted a two-year- study, funded by the FBI, into whether algorithms could be used to accurately to identify and match tattoos of inmates and others under arrest.
The project turned out to be pretty successful, but the study has brought up questions about privacy of inmates and ethics of research techniques.
Dave Maass is an investigative researcher for the Electronic Frontier Foundation and has been writing about this issue. He spoke to A Martinez about the tattoo recognition technology.
On how the study works:
The NIST which is one of the oldest research institutions in the federal government has been running a research program with the FBI to encourage the development of algorithms that can identify tattoos, start to reveal their meanings, and connect people with similar tattoos to one another. This involved collecting data sets of tattoos—the first one was more than 15,000 images—and distributing them to third parties, including eight private companies. There were at least 19 third parties in all that received the data set and they were assigned a series of tests and then they either self-reported back. In a lot of cases, some didn't report back at all. ...It's a database that's already used in investigations by the FBI and various law enforcement. This dataset is mean to test and train tattoo recognition algorithms.
On the objections to the program:
When we're talking about science, there are certain ethical standards for how things should be conducted. In this case, they used—almost entirely—tattoos collected in law enforcement settings, particularly jails and prisons. While you might think that prisoners lose a lot of rights once they're in prison, they actually have a lot more rights when it comes to being subjects in scientific experiments. In this case, NIST did not go through the ethical process prior to conducting the research and as a result, a lot of the research is ethically questionable and threatens privacy and civil liberties. Part of that is these images were handed to a lot of third parties with very little restrictions, but also a lot of the tests involved some questionable methodology that seemed to target people based on their religion. A lot of the white papers and presentations and instructions that were involved talked specifically about how tattoos can reveal people's religious and ritualistic beliefs, their political ideology, and their affiliations with various groups.
On what makes tattoos different from fingerprints or photos [in investigations]:
In a lot of ways, it's very similar and there's a lot of concerns over how facial recognition is being used by the private sector and by police, and there's a lot of comparisons that can be made on how people can be tracked by their faces, or in this case, by their tattoos on CCTV, in images online, but what's different about tattoos rather than face and fingerprints is that basically fingerprints don't tell you about people's beliefs. It isn't necessarily speech but tattoos are speech. They are how we express who we are, they may show our past, things that we used to believe in, but they also show whether you're connected to a military group, connected to a labor union. You might get tattoos that show your political beliefs...but also what religion you're a part of.
On what's wrong if this helps investigations:
That might be true, but I think you can make the same argument about eliminating the Fourth Amendment [the right to be protected against "unreasonable search and seizure"] altogether. I think that, certainly, if you put video cameras in everyone's homes, maybe that would result in more successful investigations, but that isn't consistent with how Americans cherish their privacy and their freedoms.
NIST sent a statement in response to the EFF report. It's here in full:
NIST has a long history of cutting edge research and evaluations in the field of image processing. Several of these projects have assisted law enforcement agencies in advancing pattern analysis in fields such as fingerprint and ballistics matching and facial recognition. In 2014, NIST began working on a project to evaluate automated, image-based tattoo recognition technologies.
The project uses a database of tattoo images to fairly and reproducibly assess which algorithms from companies and research groups produce the highest-quality matches. The goal of the NIST project is to help ensure tattoo matching technologies are evaluated using sound science to improve accuracy and minimize mismatches. NIST communications materials (e.g., presentations, reports) have been updated to more clearly reflect the intent of this project.
NIST’s work on this project does not involve the use of human subjects as defined by federal regulations. The database contains images only, with no accompanying information on the individuals whose tattoos were photographed.
The project has been reviewed and determined to not meet the criteria for human subjects research as defined by federal regulations.
The NIST project is about measuring the effectiveness of algorithms for accurately matching digital images. The NIST project is not about the many complex law enforcement policies or approaches that may be related to images of tattoos.
We are reviewing the EFF report and will carefully consider their concerns.
Please click on the Blue Player above to hear the entire interview