Facebook: Your Face Belongs to Us
The company it will never sell users’ data, but privacy advocates says it’s uniquely aggressive in opposing any limits on its increasingly intrusive facial recognition technology.
By Jared Bennett, Center for Public Integrity
When Chicago resident Carlo Licata joined Facebook in 2009, he did what the 390 million other users of the world’s largest social network had already done: He posted photos of himself and friends, tagging the images with names.
But what Licata, now 34, didn’t know was that every time he was tagged, Facebook stored his digitized face in its growing database.
Angered this was done without his knowledge, Licata sued Facebook in 2015 as part of a class action lawsuit filed in Illinois state court accusing the company of violating a one-of-a-kind Illinois law that prohibits collection of biometric data without permission. The suit is ongoing.
Facebook denied the charges, arguing the law doesn’t apply to it. But behind the scenes, the social network giant is working feverishly to prevent other states from enacting a law like the one in Illinois.
Since the suit was filed, Facebook has stepped up its state lobbying, according to records and interviews with lawmakers. But rather than wading into policy fights itself, Facebook has turned to lower-profile trade groups such as the Internet Association, based in Washington, D.C., and the Illinois-based trade association CompTIA to head off bills that would give users more control over how their likenesses are used or whom they can be sold to.
That effort is part of a wider agenda. Tech companies, whose business model is based on collecting data about its users and using it to sell ads, frequently oppose consumer privacy legislation. But privacy advocates say Facebook is uniquely aggressive in opposing all forms of regulation on its technology.
And the strategy has been working. Bills that would have created new consumer data protections for facial recognition were proposed in at least five states this year—Washington, Montana, New Hampshire, Connecticut, and Alaska—but all failed, except the Washington bill, which passed only after its scope was limited.
No federal law regulates how companies use biometric privacy or facial recognition, and no lawmaker has ever introduced a bill to do so. That prompted the Government Accountability Office to conclude in 2015 that the “privacy issues that have been raised by facial recognition technology serve as yet another example of the need to adapt federal privacy law to reflect new technologies.” Congress did, however, roll back privacy protections in March by allowing internet providers to sell browser data without the consumer’s permission.
Facebook says on its website it won’t ever sell users’ data, but the company is poised to cash in on facial recognition in other ways. The market for facial recognition is forecast to grow to $9.6 billion by 2022, according to analysts at Allied Market Research, as companies look for ways to authenticate and recognize repeat customers in stores, or offer specific ads based on a customer’s gender or age.
Facebook is working on advanced recognition technology that would put names to faces even if they are obscured and identify people by their clothing and posture. Facebook has filed patents for technology allowing Facebook to tailor ads based on users’ facial expressions.
But despite the relative lack of regulation, the technology appears to be worrying politicians on both sides of the aisle, and privacy advocates too. During a hearing of the House Government Oversight Committee in March, Chairman Jason Chaffetz (R-UT), who left Congress on June 30, warned facial recognition “can be used in a way that chills free speech and free association by targeting people attending certain political meetings, protests, churches or other types of places in public.”
Even one of the inventors of facial recognition is worried. “It pains me to see a technology that I helped invent being used in a way that is not what I had in mind in respect to privacy,” said Joseph Atick, who helped develop facial recognition in the 1990s at Rockefeller University in New York City.
Atick, now an industry consultant, is concerned that companies such as Facebook will use the technology to identify individuals in public spaces without their knowledge or permission.
“I can no longer count on being an anonymous person,” he said, “when I’m walking down the street.”
Atick calls for federal regulations to protect people’s privacy, because without it Americans are left with “a myriad of state laws,” he said. “And state laws can be more easily manipulated by commercial interests.”
Facial recognition is here
Facial recognition’s use is increasing. Retailers employ it to identify shoplifters, and bankers want to use it to secure bank accounts at ATMs. The Internet of things—connecting thousands of everyday personal objects from light bulbs to cars—may use an individual’s face to allow access to household devices. Churches already use facial recognition to track attendance at services.
Government is relying on it as well. President Donald Trump staffed the U.S. Homeland Security Department transition team with at least four executives tied to facial recognition firms. Law enforcement agencies run facial recognition programs using mug shots and driver’s license photos to identify suspects. About half of adult Americans are included in a facial recognition database maintained by law enforcement, estimates the Center on Privacy & Technology at Georgetown University Law School.
To tap into this booming business, companies need something only Facebook has—a massive database of faces.
Facebook now has 2 billion monthly users who upload about 350 million photos every day—a “practically infinite” amount of data that Facebook can use to train its facial recognition software, according to a 2014 presentation by an engineer working on DeepFace, Facebook’s in-house facial recognition project.
“When we invented face recognition, there was no database,” Atick said. Facebook has “a system that could recognize the entire population of the Earth.”
Facebook says it doesn’t have any plans to directly sell its database. “We do not sell people’s facial recognition template or make them available for use by developers or advertisers, and we have no plans to do so,” Facebook spokesman Andy Stone said in an email.
But Facebook currently uses facial recognition to organize photos and to support its research into artificial intelligence, which Facebook hopes will lead to new platforms to place more focused targeted ads, according to public announcements made by the company. The more Facebook can recognize what is in users’ photographs using artificial intelligence, the more the company can learn about users’ hobbies, preferences, and interests—valuable information for companies looking to pinpoint sales efforts.
For example, if Facebook identifies a user’s face and her friends hiking in a photo, it can use that information to place ads for hiking equipment on her Facebook page, said Larry Ponemon, founder of the Ponemon Institute, a privacy and security research and consulting group.
“The whole Facebook model is a commercial model,” Ponemon said, “gathering information about people and then basically selling them products” based on that information.
Facebook hasn’t been consistent about what it plans to do with its facial data. In 2012, at a hearing of the Senate Judiciary Subcommittee on Privacy, Technology, and the Law, then-Chairman Al Franken (D-MN) asked Facebook’s then-manager of privacy and public policy, Rob Sherman, to assure users the company wouldn’t share its faceprint database with third parties. Sherman declined.
“It’s difficult to know in the future what Facebook will look like five or 10 years down the road, and so it’s hard to respond to that hypothetical,” Sherman said.
And in 2013, Facebook Chief Privacy Officer Erin Egan told Reuters, “Can I say that we will never use facial recognition technology for any other purposes [other than suggesting who to tag in photos]? Absolutely not.” Egan added, though, that if Facebook did use the technology for other purposes, the firm would give users control over it.
Nearly a decade ago, when facial recognition was still in its infancy, Illinois passed the Biometric Information and Privacy Act of 2008, after a fingerprint-scanning company went bankrupt, putting the security of the biometric data the company collected in doubt.
“The Illinois law is a very stringent law,” said Chad Marlow, policy counsel at the American Civil Liberties Union. “But it’s not inherently an unreasonable law. Illinois wanted to protect its citizens from facial recognition technologies online.”
That may include, possibly, Facebook’s Tag Suggestions application. First introduced in 2010, Tag Suggestions allows Facebook users to label friends and family members in photos with their name using facial recognition. When a user tags a friend in a photo or selects a profile picture, Tag Suggestions creates a personal data profile that it uses to identify that person in other photos on Facebook or in newly uploaded images.
Facebook started quietly enrolling users in Tag Suggestions in 2010 without informing them or obtaining their permission. By June 2011, Facebook announced it had enrolled all users, except for a few countries.
That’s what upset Licata, who works in finance in Chicago. In the lawsuit against Facebook, which names two other plaintiffs, Licata alleges that every time he was tagged in an image or selected a new profile picture, Facebook “extracted from those photographs a unique faceprint or ‘template’ for him containing his biometric identifiers, including his facial geometry, and identified who he was,” according to the lawsuit. “Facebook subsequently stored Licata’s biometric identifiers in its databases.”
The other plaintiffs also claim that by using their data to build DeepFace, Facebook deprived them of the monetary value of their biometric data. The statute carries penalties up to $5,000 per violation, which potentially could include thousands of Illinois residents.
Licata declined an interview request through the law firm representing him, Chicago-based Edelson PC, which specializes in suing technology companies over privacy violations. The firm’s founder, Jay Edelson, is a controversial figure. Some technologists and colleagues view him as an opportunist—a “leech tarted up as a freedom fighter”—according to a New York Times profile.
Facebook declined the Center for Public Integrity’s requests to comment on the lawsuit specifically but said in an email that “our work demonstrates our commitment to protecting the over 210 million Americans who use our service.” Facebook told The New York Times in 2015 that the BIPA lawsuit “is without merit, and we will defend ourselves vigorously.”
Facebook says users can turn off Tag Suggestions, but critics say the process is complex, making it likely the feature will remain active.
And many Facebook users don’t even know data about their likenesses are being stored. “As a person who has been tagged, there should be some agreement at least that this is acceptable” before Facebook enrolls users in Tag Suggestions, said privacy researcher Ponemon. “But the train has left the station.”
In 2016, just 21 days after the judge in the Licata case ruled against a Facebook motion that the Illinois law only applies to in-person scans, not images or video, an amendment to BIPA that would have defined facial scans just that way was offered in the state Senate. After consumer groups such as the World Privacy Forum and the Illinois Public Interest Research Group wrote letters of opposition, the measure was withdrawn by its sponsor, state Sen. and Assistant Majority Leader Terry Link (D-Vernon Hills). Link did not respond to requests for comment.
Facebook has expressed support for the amendment but won’t confirm or deny its involvement in the attempt. The effort fits a pattern, said Alvaro Bedoya, executive director of the Center on Privacy and Technology at Georgetown University.
“Their approach has been, ‘If you sue us, it doesn’t apply to us; if you say it does apply to us, we’ll try to change the law,’” Bedoya said. “It is only laws like Illinois’ that could put some kind of check on this authority, so it is no coincidence that [Facebook] would like to see this law undone. This is the strongest privacy law in the nation. If it goes away, that’s a big deal.”
Facebook’s hidden lobbying
Facebook started lobbying the federal government in earnest around 2011, when it reported spending nearly $1.4 million. By 2016, the amount grew more than five times, to almost $8.7 million, when Facebook lobbied on issues such as data security, consumer privacy, and tax reform, according to the Center for Responsive Politics.
Facebook spends much less to influence state lawmakers. According to reports compiled by the National Institute on Money in State Politics, it spent $670,895 on lobbying in states in 2016, a 64 percent jump from $373,388 in 2014. Facebook has an active presence in a handful of states—primarily California and New York—but it only hired its first lobbyist in Illinois for this year’s session.
Facebook prefers to work through trade associations to influence policy. Sources in the Illinois Legislature told the Center for Public Integrity that the BIPA amendment attempt, which would have redefined facial recognition, was led by CompTIA, a trade group that bills itself as “the world’s leading tech association.” CompTIA declined to comment in detail but confirmed that Facebook is among its members.
Facebook declined to comment about whether it was behind the amendment. When Edelson lawyers asked for information about Facebook’s lobbying related to BIPA, Facebook’s lawyers successfully requested the court to seal those records, keeping the information private.
On its website, Facebook says it is a member of 56 groups and 108 third-party organizations that it works with “on issues relating to technology and Internet policy.” CompTIA, despite acknowledging Facebook is a member, isn’t on the list.
At the Facebook annual shareholders meeting in Redwood City, California, last month, more than 90 percent of the shares voted were opposed to a proposal that would have required the company to provide more information about its political associations, including grass-roots lobbying.
CompTIA, which absorbed the Washington, D.C.-based tech advocacy group TechAmerica in 2015, employs one permanent lobbyist in Illinois and contracts with the Roosevelt Group, one of Illinois’ “super lobbyists,” which last year represented lobbying powerhouses AT&T Illinois, payday lender PLS Financial Services, and the influential Illinois Retail Gaming & Operators Association.
In August 2016 CompTIA published a blog post about the practical applications of biometrics, and labeled BIPA “problematic” because terms such as “consent” and “facial recognitions” are vaguely defined and it “invites an avalanche of litigation.”
CompTIA made political contributions to just two non-candidate groups in 2016—in the two states with the strictest privacy laws, Illinois and Texas, according to the National Institute of Money in State Politics. CompTIA gave $21,225 last year to the Illinois Democratic Party.
CompTIA also gave $5,000 to the Republican Party in Texas, where Republican Attorney General Ken Paxton is charged with enforcing the state’s biometric privacy regulations, according to the institute. Texas has enacted one of the stricter biometric privacy laws in the nation. Signed in 2009, the law requires companies to obtain an individual’s permission to capture a biometric identifier such as a facial image. But unlike Illinois’ law, it doesn’t allow state residents to sue and leaves the enforcement authority solely with the attorney general.
The Texas attorney general’s office declined to comment on whether it has pursued lawsuits on biometric privacy violations. There’s no indication that Paxton’s office has ever completed an investigation, according to a review of records.
‘They will descend on you’
Alaska, Connecticut, Montana, New Hampshire, and Washington proposed biometric privacy laws this past legislative session, but all failed except for a weakened version that survived in Washington. Two other states—Arizona and Missouri—proposed narrower bills that provide privacy protections just for students, but both fizzled out in committee. Illinois tabled a proposed amendment to BIPA that would have strengthened the law by barring companies from making submission of biometric data a requirement of doing business.
Facebook, along with Google Inc., Verizon Communications Inc., and trade groups like CompTIA, had a hand in blocking or weakening the biometric privacy bills in Montana, Washington, and Illinois, according to a Center for Public Integrity review.
What happened in Montana is typical. Katherine Sullivan, a small business owner and intellectual privacy lawyer turned privacy advocate, helped write a biometric privacy bill that Democratic Rep. Nate McConnell introduced this year in the Montana Legislature.
“Everyone I talked to as a citizen thought it was a good idea,” Sullivan said.
Still, Sullivan said she was warned that lobbyists representing powerful companies would come out against the law. “‘They will descend on you,’” Sullivan said she was told.
The Montana bill was introduced Feb. 17 and assigned to the House Judiciary Committee. Only one hearing on the bill was held, on Feb. 23. Lobbyists from Verizon, the Internet Coalition, which represents Internet and ecommerce companies including Facebook, and the Montana Retail Association showed up in opposition to the bill.
At the hearing, Jessie Luther, a lobbyist from Verizon, read a letter signed by CompTIA; the Internet Coalition; TechNet, a network of chief executives from technology companies; and the State Privacy and Security Coalition, a group of major internet communications, retail, and media companies. All three count Facebook as a member.
The letter, addressed to state Rep. Alan Doane, chairman of the Judiciary Committee, warned that the proposed legislation “would put Montana residents and businesses at much greater risk of fraud, as well as open the door to wasteful class action lawsuits against Montana businesses that receive biometric data.” It also warned that the bill would prevent using biometrics for “beneficial purposes” such as accessing and securing personal accounts.
Doane said in an interview he doesn’t remember the letter but agreed with many of its points. On Feb. 27, the bill was tabled in committee.
The ‘NRA approach’
Tough privacy legislation that would have prohibited the collection of biometric information without prior consent and allow individuals to sue companies that violate the law also fizzled out in New Hampshire and Alaska. A weaker bill in Connecticut that would have prohibited brick-and-mortar stores from using facial recognition for marketing purposes died in committee.
Washington’s law requires companies to obtain permission from customers before enrolling their biometric data into a database for commercial use and prohibits companies from selling, leasing, or otherwise handing the data over to a third party without consent. But it does not allow individuals to sue companies directly.
More important, some privacy advocates say, the law exempts biometric data pulled from photographs, video, or audio recordings, similar to the amendment CompTIA had lobbied for in Illinois as a way to weaken BIPA, which would exempt Facebook’s Tag Suggestions.
Earlier versions of the law won the approval of big tech companies such as Google and Microsoft Corp., and the privacy advocacy group the Electronic Frontier Foundation. But in 2016, EFF pulled its support when the bill was amended to omit “facial geometry,” which Adam Schwartz, a senior staff attorney at EFF, said would cover facial recognition.
Schwartz said the final statute is weaker than BIPA because the law’s language is written in such a way that it may allow companies to capture facial recognition data without informed notice or consent.
The statute “appears to have been tailored to protect companies that are using facial recognition,” Schwartz said.
Democratic state Rep. Jeff Morris, one of the bill’s sponsors, disagrees. Morris said the law covers any data that can be used to identify a person by unique physical characteristics, including applications that use “precise measurements between the bridge of your nose and your eyes.”
But Morris said that while most of the big tech companies such as Microsoft, Amazon, and Google supported the bill in its final form, Facebook remained opposed.
Facebook’s hired lobbyist in Washington—Alex Hur, a former aide to state Speaker of the House Frank Chopp—was “lobbying quite ferociously on the bill,” Morris said. Facebook objected to the bill, he said, because it included as protected data “behavioral biometrics,” which refers to data on how a person moves, including an individual’s gait as recorded in videos.
Hur did not respond to requests for comment.
One of the trade groups working on Facebook’s behalf in Washington was the Washington Technology Industry Association. At a hearing on the legislation in February, Jim Justin, a WTIA representative, argued tagging services like Facebook’s should be exempt from the law.
“Given facial recognition, that data should be protected,” Justin said, “but if you are tagging someone on Facebook and simply using their name, we don’t think that falls under what should be protected, given that that person provided consent.”
A CompTIA lobbyist also spoke at the February hearing, asking lawmakers to take a “limited approach” to biometric privacy.
Morris said CompTIA adopts what he calls the “NRA approach” to lobbying. “They basically say, ‘You’ll take our innovation out of our cold, dead hands,’” he said.
“This is a pretty common public-affairs tactic,” Morris added, “an association that does the dirty work so your company isn’t tarnished.”
‘Didn’t know they existed until…’
State legislatures are beginning to recognize that many personally identifying technologies may require additional regulatory attention—and technology companies such as Facebook and their trade groups are gearing up to fight them.
Lawmakers in Illinois formed a committee this year to discuss technology issues such as data privacy. The CyberSecurity, Data Analytics and IT committee in the Illinois House of Representatives held its first hearing in March.
The formation of the committee brought national attention to Springfield.
“It has brought in groups from D.C.,” like the Internet Association, said Rep. Jaime Andrade Jr. (D-Chicago), the committee’s chairman.
CompTIA also has been “very active,” he said.
“I didn’t know they existed until the committee” formed, Andrade added. “As soon as the committee was created they came in and introduced themselves.”