Privacy hawks fret as FBI upgrades biometrics capacities

Update: 5:27 PM PT The computer revolution arrived late at the FBI, which was still collecting and matching fingerprints in 1999 in much the same way that it did when the agency first began collecting the images in 1924. But that's been changing lately and privacy hawks are watching closely.

As the millennium neared, the agency finally traded in its manual system for one in which a database of fingerprints and associated criminal histories could be searched and updated. Now, the next step.

The FBI is going ahead with an advanced biometric identification service that it has been testing and will phase into operation around the nation over the next couple of years. The official launch of the so-called Next Generation Identification program is part of the agency's billion-dollar upgrade of the agency's national fingerprint database, though some states have already participated in a pilot program by already uploading their photos. (It's about 60 percent deployed.)

It's a big step for the FBI, which is hoping to take advantage of recent advances in the field to make faster arrests.

Indeed, since 1993, the error rate of face recognition systems has dropped by a factor of 272, Acquisti told a U.S. Senate subcommittee earlier this summer. But as the hit rate gets more accurate, the FBI's capabilities will inevitably raise concerns about how it uses that new technology. The FBI's Jerome Pender told the Senate in July that the tests involved only included mugshots of known criminals and that future searches would be subject to rules governing criminal justice agencies.

He testified that each participating pilot state or agency must submit written statements detailing how they will use and protect the information from unauthorized disclosure.

"Pilot participants are informed that information derived from pilot search requests and resulting responses is to be used only as an investigative lead. Results are not to be considered as positive identifications," he added.

Sounds nice in theory, but privacy groups still want to see the fine print. The FBI has declined to offer specifics about the algorithms it uses in the system. In an email response to CNET, Stephen G. Fischer Jr. said that "the FBI is tentatively planning to host a meeting of federal law enforcement and national security agencies with privacy and civil liberties groups to discuss various aspects of federal government uses of facial recognition technology later this year. Details will be forthcoming."

But in a 2010 presentation on facial recognition, the FBI's Richard Vorder Bruegge talked about the agency's desire to use facial recognition system to "identify subjects in public datasets" and "conduct automated surveillance at lookout locations."

That was a red flag for the Electronic Frontier Foundation, which is not satisfied with the FBI's privacy statement and has expressed concerns about the FBI's plans to build up its facial recognition capabilities. EFF attorney Jennifer Lynch also raised the possibility that the FBI still could also add civilian photos to the database with nobody able to prevent it.

It is impossible to tell exactly how the FBI plans to acquire and use facial recognition data now and in the future. However, given the information in these new documents and the FBI's broad goals for face recognition data, the time is right for laws that limit face recognition data collection.

Here's a summary of some of the advances that the FBI is touting:

• Interstate Photo System enhancements including the ability to accept and search for photographs of scars, marks, and tattoos.
• Better fingerprint identification technology for faster, more accurate identification processing.
• A new "Rap Back" service notifying authorized agencies about criminal activity reported "on individuals holding positions of trust."
• National palm print system that will allow the storage and search of palm print submissions from law enforcement and criminal justice agencies.
• A more powerful system based around "multimodal biometrics" that will eventually extend to a biometric understanding of things like voice, or facial recognition. Computer scientists have researched digital face recognition for the last four decades and now believe that under certain conditions, machine face recognition performance can be comparable or even better than humans at recognizing face," according to Alessandro Acquisti of Carnegie Mellon University in Pittsburgh.