Since early 2011, in
an effort to increase resources available to law enforcement (LE), the Federal
Bureau of Investigation (FBI) has incrementally rolled out their Next
Generation Identification (NGI) system which not only enhances the preexisting
Integrated Automated Fingerprint Identification System but expands the biometric
data-capture function exponentially. According to the FBI, NGI advancements
will “increase the range and quality of its identification and investigative
capabilities (Federal Bureau of Investigation,
n.d.) ”
through components such as the Advanced Fingerprint Identification Technology,
Repository for Individuals of Special Concern, Latent and Palm Prints, Rap
Back, Facial Recognition, and Iris Scan (still currently being beta-tested). NGI
additionally enables multi-national collaboration, an especially important
feature in today’s war on terror.
FBI Criminal
Justice Information Services Assistant Director Thomas E. Bush III purports NGI
is not a data collection system but a “database that stores and aggregates
information sent to it…from law enforcement agencies. (qtd. Smith, 2008)” yet Privacy
advocates suggests otherwise, especially with regards to NGI’s facial
recognition capabilities. While the accuracy of fingerprint analysis via NGI has
improved from 92% to 99.6 % (Federal
Bureau of Investigation, n.d.) , the FBI only
guarantees that facial recognition software will result in an 85% accuracy rate
within a list of 50 candidates generated and only “when the true candidate
exists” (qtd. Lynch, 2014). The minimum recommended resolution is .75
megapixels, a shockingly low threshold when compared against the newer iPhones
which possess 8-megapixel resolution (Lynch, 2014) . (Even Google Glass,
which banned the use of facial recognition applications, possess a 5-megapixel
camera! (Warman, 2013) )
Unfortunately, much
of America’s video surveillance infrastructure utilizes technology that does
“meet the necessary resolution standards to make accurate facial recognition” (Walker, 2014) . Furthermore, a
German field study in facial recognition software in 2006 demonstrated only a
60% daytime accuracy rate and a dismal 10% nighttime rate (Smith, 2008) . The FBI insists
that NGI enables a zero percent false positive feedback because the system does
not generate positive identifications but investigative leads. Yet, a database
which contains both criminal and non-criminal (i.e. background check for
employment) data, capable up processing up to 52 million photo queries a day, has
an error rate of almost 8 million people.
Even more unnerving
than the lack of accuracy remains a systemic failure to establish working
guidelines. Documents released to Electronic Frontier Foundation as a result of
a freedom of information lawsuit illuminate that “the FBI and Congress have
thus far failed to enact meaningful restrictions on what types of data can be
submitted to the system, who can access the data, and how the data can be used” (Lynch, 2014) . Many deemed this as
unacceptable particularly given that the annual financial obligation necessary
to support NGI is suspected of being approximately one billion dollars (Smith, 2008) . Nonetheless, the
FBI can mitigate concerns over privacy encroachment through transparency in
both policy and practice.
The following
proposals can alleviate, if not then placate, concerns over privacy intrusions
as well as instill the public with a sincere attempt by LE to minimize mistaken
identities:
·
instituting a comprehensive
policy, based upon significant field testing of enhanced capabilities, that includes
standards required to accept data (i.e. higher camera resolution, minimum Galton
Points, etc.), the full scope of data complied, methods of data acquisition, accessibility
requirements that include limits that cannot be exceeded without a warrant,
flow of information, data usage applications, and a recourse for data removal
(conditional);
·
the establishment
of multiple databases that separate the non-criminal, criminal, and terror
suspect information;
·
Non-introduction of
criminal data without criminal conviction (except terror investigations);
·
Alignment with both
the Privacy Act and Freedom of Information Act;
·
Congressional
oversight that involves semi-annual briefings; and
·
System accessibility
preclude any attempts for information from any non-verifiable and/or non-active
LE entities.
Reference
List
Federal Bureau of Investigation. (n.d.). Next
Generation Identification (NGI). Retrieved from fbi.gov:
https://www.fbi.gov/services/cjis/fingerprints-and-other-biometrics/ngi
Lynch, J. (2014). FBI Plans to Have 52 Million
Photos in its NGI Face Recognition Database by Next Year. Retrieved from
Electronic Frontier Foundation:
https://www.eff.org/deeplinks/2014/04/fbi-plans-have-52-million-photos-its-ngi-face-recognition-database-next-year
Smith, J. L. (2008). The FBI's Next Generation
Identification Database. Retrieved from Hendon Publishing.
Walker, L. (2014). FBI Announces Its Facial
Recognition System is Ready to Go. Retrieved from Newsweek:
http://www.newsweek.com/fbi-announces-its-facial-recognition-system-ready-go-270955
Warman, M. (2013). The Creepiest Google Glass
Feature Has Been Banned: No Facial Recognition Apps Allowed. Retrieved
from Business Insider:
http://www.businessinsider.com/google-glass-facial-rec-apps-banned-2013-6
No comments:
Post a Comment