The FBI's Next Generation Biometrics: Bigger and Faster but Much Worse for Privacy
Last week, the Center for Constitutional Rights (CCR) and several other organizations released documents from a FOIA lawsuit that expose the concerted efforts of the FBI and DHS to build a massive database of personal and biometric information. This database, called “Next Generation Identification” (NGI), has been in the works for several years now. However, the documents CCR posted show for the first time how FBI has taken advantage of the DHS Secure Communities program and both DHS and the State Department’s civil biometric data collection programs to build out this $1 billion database.
Unlike some government initiatives, NGI has not been a secret program. The FBI brags about it on its website (describing NGI as “bigger, faster, and better”), and both DHS and FBI have, over the past 10+ years, slowly and carefully laid the groundwork for extensive data sharing and database interoperability through publicly-available privacy impact assessments and other records. However, the fact that NGI is not secret does not make it OK. Currently, the FBI and DHS have separate databases (called IAFIS and IDENT, respectively) that each have the capacity to store an extensive amount of information—including names, addresses, social security numbers, telephone numbers, e-mail addresses, fingerprints, booking photos, unique identifying numbers, gender, race, and date of birth. Within the last few years, DHS and FBI have made their data easily searchable between the agencies. However, both databases remained independent, and were only “unimodal,” meaning they only had one biometric means of identifying someone—usually a fingerprint.
In contrast, as CCR’s FOIA documents reveal, FBI’s NGI database will be populated with data from both FBI and DHS records. Further, NGI will be “multimodal.” This means NGI is designed to allow the collection and storage of the now-standard 10-print fingerprint scan in addition to iris scans, palm prints, and voice data. It is also designed to expand to include other biometric identifiers in the future. NGI will also allow much greater storage of photos, including crime scene security camera photos, and, with its facial recognition and sophisticated search capabilities, it will have the “increased ability to locate potentially related photos (and other records associated with the photos) that might not otherwise be discovered as quickly or efficiently, or might never be discovered at all.”
The FBI does not just collect and store data from people caught up in the criminal justice system; about 1/3 of the data collected and reviewed in IAFIS is from civil sources such as attorney bar applications, federal and state employees, and people who work with children or the elderly. In the past, the FBI has not allowed these records to include photos and has segregated civil records from criminal data. Civil records were also not included in bulk checks for criminal investigative purposes. NGI may take down these barriers, however. There is some evidence to show the FBI is considering including this data in future NGI database searches and, according to the CCR FOIA documents, has already begun to include civil records from DHS and State Department database files such as visa applications, immigration records, and border entries and exits.
So why should we be worried about a program like NGI, which the FBI argues will “reduce terrorist and criminal activities”? Well, the first reason is the sheer size of the database. Both DHS and FBI claim that their current biometrics databases (IDENT and IAFIS, respectively) are the each the “largest biometric database in the world.” IAFIS contains 66 million criminal records and 25 million civil records, while IDENT has over 91 million individual fingerprint records.
Once these records are combined into one database and once that database becomes multimodal, as we discussed in our 2003 white paper on biometrics, there are several additional reasons for concern. Three of the biggest are the expanded linking and tracking capabilities associated with robust and standardized biometrics collection systems and the potential for data compromise.
Already, the National Institute for Standards and Technology, along with other standards setting bodies, has developed standards for the exchange of biometric data. FBI, DHS and DoD’s current fingerprint databases are interoperable, indicating their systems have been designed (or re-designed) to read each others’ data. NGI will most certainly improve on this standardization. While this is good if you want to check to see if someone applying for a visa is a criminal, it has the potential to be very bad for society. Once data is standardized, it becomes much easier to use as a linking identifier, not just in interactions with the government but also across disparate databases and throughout society. This could mean that instead of being asked for your social security number the next time you apply for insurance, see your doctor, or fill out an apartment rental application, you could be asked for your thumbprint or your iris scan.
This is a big problem if your records are ever compromised because you can’t change your biometric information like you can a unique identifying number such as an SSN. And the many recent security breaches show that we can never fully protect against these kinds of data losses.
The third reason for concern is at the heart of much of our work at EFF. Once the collection of biometrics becomes standardized, it becomes much easier to locate and track someone across all aspects of their life. As we said in 2003, “EFF believes that perfect tracking is inimical to a free society. A society in which everyone's actions are tracked is not, in principle, free. It may be a livable society, but would not be our society.”
Unfortunately, biometric data collection is not limited to NGI or even to the legacy DHS, FBI and DoD fingerprint collection programs. The federal government and states have been steadily expanding their DNA collection efforts over the last 10 years as well. Currently all 50 states, the federal government and the District of Columbia collect and share DNA records through the FBI’s CODIS database. At least 15 of those states, as of 2010, collect DNA from defendants convicted of misdemeanor offenses. And as of 2009, under the federal DNA Fingerprint Act of 2005 and several recently-expanded state statutes, at least 21 states and the federal government collect DNA samples from any adult arrested for (not just convicted of) a crime. This has led to an exponential increase in the amount of DNA collected in the United States on an annual basis, with nearly 1.7 million samples processed (pdf p. 8) in 2009, alone. As of 2011, the National DNA Index or NDIS (the federal level of CODIS) contains over 9,748,870 offender profiles, and the states’ individual databases are each expanding as well.
Currently, it doesn’t appear the FBI plans to incorporate the DNA data held by CODIS into NGI. However, NGI has been designed to be flexible and to be able to incorporate additional biometric identifiers as the need arises in the future. This means that we can’t rule anything out. FBI claims NGI “doesn’t threaten individual privacy,” but the government’s continuing efforts to collect, store and track the biometric data for so many Americans and foreigners cannot bode well for a society that values privacy.
© 2011 Electronic Frontiers Foundation