As you scan the face on that giant billboard, it may just be scanning your face right back.
Increasingly sophisticated digital facial-recognition technology is opening new possibilities in business, marketing, advertising and law enforcement while exacerbating fears about the loss of privacy and the violation of civil liberties.
Businesses foresee a day when signs and billboards with face-recognition technology can instantly scan your face and track what other ads you’ve seen recently, adjust their message to your tastes and buying history and even track your birthday or recent home purchase. The FBI and other U.S. law enforcement agencies already are exploring facial-recognition tools to track suspects, quickly single out dangerous people in a crowd or match a grainy security-camera image against a vast database to look for matches.
Many fear that future is coming too quickly, with facial-recognition technology becoming increasingly advanced, available and affordable before restrictions on its use can be put into place. Concerns have been raised on Capitol Hill in recent weeks that FBI searches using the technology could trample Fourth Amendment protections against unreasonable search and seizure, while some in the industry say excessive regulations could cripple cutting-edge technology.
“In our country, government shouldn’t be looking over your shoulder unless it has a reason,” said Jay Stanley, senior policy analyst with the American Civil Liberties Union’s speech, privacy and technology project. “They should not be collecting data on innocent subjects.”
The potential to “data-mine” raw video or photography using facial-recognition technology is another concern, he said, but one that could clash with First Amendment rights on the right to photograph.
Facebook enters the fray
Sparking fresh concerns on the commercial front was social media behemoth Facebook’s acquisition last month of the Israeli technology company Face.com.
The acquisition enabled Facebook to implement a feature called “tag suggestions,” which lets the social networking site use facial recognition technology to make suggestions about who is in a picture. Users can opt out of this service, but the default setting is “on.”
Although there are no rules governing the use of this information, Facebook, which is estimated to have pictures and demographic information for more than 900 million users, has publicly stated multiple times that it will not allow any third-party access to its database. This includes the government, though it has cooperated with law enforcement officials in a limited manner, a company spokesman said at a July 18 Senate hearing on face-recognition technology concerns.
One proposal is for a comprehensive privacy law, similar to those in many European countries. This would provide for a “privacy commissioner” to deal with concerns raised by technological developments as well as basic rules such as a requirement to ask permission before using any pictures or videos.
“If companies take and use [my picture], that’s fine,” said Justin Brookman, director for consumer privacy at the Center for Democracy and Technology. “Just tell me about it.”
But Marcus Dunn, director of government relations at the Security Industry Association, said the technology is not advanced to the point where new legislation is needed, and new regulations could end up doing more harm than good.
Before then, he said, new, cheaper or less-invasive technology may supersede facial-recognition tools, so the government should proceed with caution.
“If taxpayers don’t have to pay for a whole new agency to be set up for it, or for three extra staff people, or for a new bureau at the FBI, that’s savings for taxpayers, and that’s always good,” he said.
Work in progress
Despite major advances in the technology in recent years, many limitations remain, said Brian Martin, director of biometric research for MorphoTrust USA Inc. He told a Senate hearing this month that while recognition algorithms match faces accurately 99.7 percent of the time, the studies are conducted only in controlled situations with perfect pictures and perfect lighting. In less-pristine settings, pictures may be blurry or have large shadows, and the reliability rates will suffer.
The success rate also represents only verification, rather than identifying an unknown person from a picture. Such identifications almost always need a human observer to verify that the computer is correct, he said.
“I don’t think that the accuracy of face recognition, for good-quality pictures, will continue to improve at the rate that it has in the last 10 years,” Mr. Martin said. “However, for the uncontrolled cases, when you’re not looking at the camera, I do think that over the next couple of decades there will be a substantial improvement in accuracy to help these forensic type of cases.”
The FBI’s Next Generation Identification program also may accelerate the rate of progress. It will provide a national database of mug shots, enabling law enforcement officials to use the facial-recognition technology to quickly search pictures of suspects against photos of anyone who has been arrested. Set to take effect in 2014, it has caused concerns that officials can discover a criminal past of anyone for whom they can obtain a picture, with or without probable cause.
Sen. Al Franken, the Minnesota Democrat who chaired the Senate hearing, noted that FBI training manuals already show facial-recognition technology being used to identify protesters.
“I fear the FBI pilot program could be abused not only to identify protesters but to target them as well,” he said.
Larry Amerson, president of the National Sheriffs’ Association and sheriff of Calhoun County, Ala., told lawmakers that one major use for facial-recognition technology would be to prevent the accidental release of felons who still have time to serve. Prisoners set to be released often will switch identities with someone else for a variety of reasons, enabling the wrong person to go free. Georgia has used facial-recognition technology for eight years to prevent this from happening, Mr. Amerson said.
“While fingerprints take hours and days for analysis, some advanced facial recognition in use today by U.S. law enforcement is as accurate as fingerprints, but results are obtained in seconds, not hours, in identifying criminals and perpetrators attempting to use false identities and aliases,” Mr. Amerson said.
Some go further, arguing that there is nothing wrong with using the technology to aid criminal investigations. Nita A. Farahany, a law and genome sciences and policy professor at Duke University, said using a computer to scan faces is the same as a police officer scanning faces in a crowd.
“A Fourth Amendment search only occurs when the government intrudes upon a legally cognizable interest of an individual,” Ms. Farahany told the hearing. “Neither scanning an individual’s face in public afar, nor querying a database to see if there is a match, intrudes on a cognizable privacy interest of an individual.”
Regardless of whether the technology violates rights or not, legislation is necessary, the Center for Democracy and Technology’s Mr. Brookman argued.
“Something has to be done, because otherwise we are living in a world of ubiquitous identity” where you can’t walk out your front door,” he said.