Clearview AI Offered Thousands Of Cops Free Trials

A controversial facial recognition tool designed for policing has been quietly deployed across the country with little to no public oversight. According to reporting and data reviewed by BuzzFeed News, more than 7,000 individuals from nearly 2,000 public agencies nationwide have used Clearview AI to search through millions of Americans’ faces, looking for people, including Black Lives Matter protesters, Capitol insurrectionists, petty criminals, and their own friends and family members.

BuzzFeed News has developed a searchable table of 1,803 publicly funded agencies whose employees are listed in the data as having used or tested the controversial policing tool before February 2020. These include local and state police, US Immigration and Customs Enforcement, the Air Force, state healthcare organizations, offices of state attorneys general, and even public schools.

In many cases, leaders at these agencies were unaware that employees were using the tool; five said they would pause or ban its use in response to questions about it.

Our reporting is based on data that describes facial recognition searches conducted on Clearview AI between 2018 and February 2020, as well as tens of thousands of pages of public records, and outreach to every one of the hundreds of taxpayer-funded agencies included in the dataset.

The data, provided by a source who declined to be named for fear of retribution, has limitations. When asked about it in March of this year, Clearview AI did not confirm or dispute its authenticity. Some 335 public entities in the dataset confirmed to BuzzFeed News that their employees had tested or worked with the software, while 210 organizations denied any use. Most entities — 1,161 — did not respond to questions about whether they had used it.

Still, the data indicates that Clearview has broadly distributed its facial recognition software to federal agencies and police departments nationwide, offering the app to thousands of police officers and government employees, who at times used it without training or oversight. Often, agencies that acknowledged their employees had used the software confirmed it happened without the knowledge of their superiors, let alone the public they serve.

Such widespread use of Clearview means that facial recognition may have been used in your hometown with very few people knowing about it.

In a statement to BuzzFeed News, Hoan Ton-That, the company’s cofounder and CEO, said it was “gratifying to see how quickly Clearview AI has been embraced by US law enforcement.” He declined to answer more than 50 detailed questions about the company’s practices and relationships with law enforcement agencies.

“Americans shouldn’t have to rely on BuzzFeed to learn their local law enforcement agencies were using flawed facial recognition technology,” Sen. Ron Wyden, an Oregon Democrat, told BuzzFeed News. “This report pulls back the curtain on Clearview’s shady campaign to encourage the secret adoption of its service by local police. Inaccurate facial recognition will inevitably cause innocent people to be wrongly accused and convicted of crimes and could very well lead to tragedies.”

For years, law enforcement agencies have experimented with facial recognition, a technology that promises to help identify people of interest by matching surveillance photos to known images — such as a headshot from a driver’s license or passport. But there are several barriers to its adoption, including high costs, unreliable results, and public opposition.

Clearview has pushed its technology into the mainstream with a product it claims is both more accurate and cost-effective than those of its competitors. For a time, it made the tool accessible via a free trial to almost any law enforcement officer who wanted to sign up.

“I found this site on a law enforcement web site last year, I set up an account to see if it worked,” Adrian Williams, police chief in Wilson’s Mills, North Carolina, wrote to BuzzFeed News in December 2020.

If you have information about Clearview AI, or other facial recognition technology used by law enforcement, please email us at tips@buzzfeed.com. Or, to reach us securely, see this page.

Williams said he had tested Clearview AI using his own personal photos, but the software returned no matches. “I ran two known persons to see if they came back with any useful info. I didn’t think it worked the way the ad said it would,” he said.

The New York City–based startup claims to have amassed one of the largest-known repositories of pictures of people’s faces — a database of more than 3 billion images scraped without permission from places such as Facebook, Instagram, and LinkedIn. If you’ve posted images online, your social media profile picture, vacation snapshots, or family photos may well be part of a facial recognition dragnet that’s been tested or used by law enforcement agencies across the country.

Data analyzed by BuzzFeed News indicated that individuals at some 1,803 agencies — which were all contacted and asked whether they had ever used Clearview — ran almost 340,000 searches. Based on conversations with people who have used the software, a Clearview search may include a demonstration scan at a trade show, a police officer looking up a colleague as a test, or an actual investigative attempt to find a person of interest. While this database represents only a snapshot of Clearview’s reach as of February 2020, it provides unprecedented insight into how facial recognition technology can spread through law enforcement agencies across the country.

Smaller police departments were among Clearview’s earliest users. Officials in Mountain Brook, Alabama, which has a population of about 20,000, tested the product and ran nearly 590 searches. In Illinois, the secretary of state’s office ran nearly 8,900 searches, telling BuzzFeed News that the software was used “to assist other law enforcement agencies in their investigations.”

A police department crime analyst in Leawood, Kansas, tried Clearview thanks to a recommendation from a peer, only to discover another detective had already been using the software, a spokesperson said. The two officers ran 50 searches before the department decided it didn’t want to buy the product. The Wyoming Highway Patrol gave Clearview a go for about three weeks last February, Sgt. Jeremy Beck said.

Even representatives at Washington’s Department of Fish and Wildlife and Minnesota’s Commerce Fraud Bureau confirmed to BuzzFeed News that individuals within their organizations tested the software. A spokesperson for Iowa’s public defender’s office said people used it “to better understand how law enforcement might use it.”

Following an inquiry from BuzzFeed News, the Washington National Guard’s Counterdrug Program found that an instructor at its Western Regional Counterdrug Training Center had not only registered to use Clearview’s facial recognition software without permission but had also incorporated it into the curriculum for an officer training course. This course was open to “federal, state, local and tribal law enforcement,” per emails obtained by BuzzFeed News through a public records request.

“Any mention of this software has now been stripped from our training materials,” said Karina Shagren, communications director for Washington’s military department. “And we’re now conducting a top-down review of all training programs.”

Clearview’s promotional materials and communications with law enforcement agencies were obtained via more than 140 public records requests from BuzzFeed News. The documents detail the startup’s flood-the-market strategy, in which it hawked free trials of its technology to seemingly anyone with an email address associated with the government or a law enforcement agency and told them to “run wild.” The data shows that this marketing push put an unproven and federally unregulated facial recognition tool into the hands of people associated with taxpayer-funded agencies in the District of Columbia, the US Virgin Islands, and every state except Vermont.

After March 2020, according to emails obtained via a public records request, Clearview placed a few checks on its free trial program, including the requirement of a superior’s approval and appointment of an administrator to monitor use. Clearview’s website was also updated last month to state that officers have to provide a case number before conducting a search.

But between summer 2019 and February 2020, none of these checks existed. Any officer could sign up, and Clearview explicitly encouraged them to test its software on friends and family members.

“No strings attached,” a November 2019 email to police in Lakeland, Florida, reads. “It only takes one minute to install and you can start searching immediately.”

In a statement to BuzzFeed News, Ton-That said that in two years the company had helped “thousands of agencies” solve crimes including “child exploitation, financial fraud, and murder,” but did not provide specific examples when asked.

“As a young startup, we’re proud of our record of accomplishment and will continue to refine our technology, cybersecurity, and compliance protocols,” he said. “We also look forward to working with policymakers on best practices to forge a proper balance between privacy and security that serves the interests of families and communities across America.”

Created by Ton-That, an Australian-born college dropout who traveled in far-right circles, Clearview AI debuted in 2017 as SmartCheckr, a tool for tracking people across disparate social media platforms. With funding from Facebook board member Peter Thiel, the company changed its name a year later when it began focusing on facial recognition.

Clearview has touted its software as the “world’s best facial-recognition technology,” but its most novel innovation is doing what no other company has been willing to do: rip billions of personal photos from social media and the web without permission or consent.

In his statement, Ton-That compared Clearview to a search engine and said the facial recognition tool only searches public information available on the internet. “The work we do is fully protected by the First Amendment and complies with all applicable laws,” he said.

Critics, including the American Civil Liberties Union, disagree.

“Protecting privacy means maintaining control of private information that is most revealing of our identities, our activities, our selfhood, and Clearview’s business is based on taking that control away from people without their consent,” Nathan Freed Wessler, a senior staff attorney with the ACLU, told BuzzFeed News. “You can change your Social Security number if it is taken without your consent, but you can’t change your face.”

Last May, the ACLU sued Clearview for allegedly violating an Illinois law overseeing the collection of biometric data by private companies. The company is also facing multidistrict litigation for the sale of its technology in Illinois — an alleged violation of a state biometric privacy law — a suit from the Vermont attorney general, and a suit from Latinx and Chicanx rights group Mijente. Facebook, LinkedIn, Google, and Twitter have all sent Clearview cease-and-desist letters alleging that the company violated its terms of service by scraping people’s data. (All four companies declined to say if they have plans to take further legal action against Clearview.)

None of this has hampered Clearview’s aggressive marketing efforts toward law enforcement. The company regularly promoted itself to email lists of officers with claims that its software can scan “over 1 billion faces in less than a second” and that it is “100% accurate across all demographic groups.”

This strategy has worked brilliantly. By February 2020, almost 2,000 taxpayer-funded entities and police departments across the US had at least one person run at least one search with Clearview, according to data reviewed by BuzzFeed News. The company assured those users in marketing emails that the more searches they ran, the more likely they were to match a suspect.

Wessler condemned the company’s marketing.

“Their claims of near-perfect identification have never been substantiated, and the pervasive tracking of our faces and whereabouts with a flawed technology is just too dangerous to have in the hands of the government,” he said.

In recent years, Clearview reached hundreds of law enforcement agencies by using a sales strategy commonplace among software companies like Slack or Dropbox. Rather than rely on standard procurement channels to make a sale, it also targeted individual employees with free trials. That allowed Clearview to create internal demand in a bottom-up manner with the hope that users would advocate within their departments to move to a paid version.

These free trials have helped Clearview create a broad swath of connections with local police officers; Ton-That recently claimed that 3,100 law enforcement agencies have used the software as of March 2021. Listed in the data BuzzFeed reviewed, for example, were more than 40 individuals in the New York Police Department who had collectively run over 11,000 searches — the most of any entity as of February 2020. The NYPD announced new facial recognition policies in March 2020 following a BuzzFeed News story that detailed how it used the software.

The New York State Police ran more than 5,100 searches and used the software “to generate potential leads in criminal investigations as well as homeland security cases involving a clearly identified public safety issue,” according to Beau Duffy, director of public information. The data lists other police departments that have run more than 4,000 searches as of February 2020: Bergen County Sheriff’s Office in New Jersey (more than 7,800), Indiana State Police (more than 5,700), Broward County Sheriff’s Office in Florida (more than 6,300), and Jefferson Parish Sheriff’s Office in Louisiana (nearly 4,200). Indiana State Police confirmed its use of Clearview. The Broward County and Jefferson Parish offices did not respond to multiple requests for comment. A Bergen County Sheriff spokesperson denied the agency used Clearview.

According to data reviewed by BuzzFeed News, individuals at 15 different state attorneys general offices tried the software, including those in Texas, Alabama, and New Jersey, which banned its own law enforcement agencies from using it in January 2020. The Texas attorney general’s office did not respond to multiple requests for comment, while a spokesperson for the Alabama attorney general’s office said it “had no contracts with Clearview AI.”

When BuzzFeed News notified the California attorney general’s office of accounts tied to its employees, a spokesperson said the state’s Department of Justice had “not authorized the use of facial recognition software.” The department subsequently blocked access to Clearview’s website on employees’ devices, the spokesperson added, “out of an abundance of caution.”

Individuals associated with public schools have also apparently tried the company’s facial recognition software. As of early 2020, data reviewed by BuzzFeed News lists 31 public community colleges and universities — including the University of Alabama and West Virginia University, neither of which responded to multiple requests for comment. A spokesperson for California State University, Long Beach, confirmed a detective had reviewed the platform’s capabilities on a 30-day trial.

Records seen by BuzzFeed News show that individuals associated with two high schools — Somerset Berkley Regional High School in Massachusetts and Central Montco Technical High School in Pennsylvania — and Texas’s Katy Independent School District appear to have run searches.

Officials at Central Montco Technical High School did not respond to requests for comment. A spokesperson for the Somerset Police Department, which is tied to Somerset Berkley Regional High, according to data seen by BuzzFeed News, confirmed that a detective had used Clearview on a 30-day trial. A spokesperson for Katy Independent School District’s police department said that the agency does not use facial recognition software and did not answer follow-up questions.

In marketing statements, Clearview claims its software has been used to help identify child predators. And to some extent these claims were born out in BuzzFeed News’ reporting.

ICE — as well as police in Mount Pleasant, Wisconsin, and Raleigh, North Carolina — told BuzzFeed News the software had been used in such cases. “Clearview AI is most often used to investigate formidable crimes that are extraordinary in nature, such as reports of human trafficking and shootings,” said Laura Hourigan, a spokesperson for the Raleigh Police Department. “These searches are fairly narrow in their scope, are limited, and are focused specifically on what they are looking for at that time.”

Ton-That told BuzzFeed News that law enforcement agencies have also used Clearview to identify insurrectionists who stormed the US Capitol on Jan. 6, though he declined to provide specifics. Data seen by BuzzFeed News shows that the US Capitol Police used Clearview to run more than 60 searches as of early 2020.

It’s unclear if the Capitol Police, the main law enforcement agency tasked with protecting Congress, still has access to the facial recognition tool. A spokesperson did not respond to multiple requests for comment.

Clearview’s free trials, most of which give users unlimited searches for 30 days, may have also helped put its software into the hands of employees at many of the nation’s largest law enforcement agencies. Individuals associated with the departments of Justice, Defense, and State have all apparently tried the facial recognition software, according to data reviewed by BuzzFeed News. Spokespeople for those departments declined or did not respond to requests for comment.

The data lists people at the five largest branches of the US military — the Army, Air Force, Navy, Coast Guard, and Marines — as having used the company’s software. The US Army Criminal Investigation Command, which pursues violations of military law, had run more than 1,300 searches as of February 2020, according to the data. In December 2019, the Air Force signed a $50,000 exploratory contract with Clearview “to determine if there was an operational need,” according to a branch spokesperson. The Army, Navy, and Air Force did not respond to multiple requests for comment. A spokesperson for the Marine Corps said there was no indication that anybody at the service branch had used Clearview. A spokesperson said the Coast Guard “does not use Clearview AI.”

If you have information about Clearview AI, or other facial recognition technology used by law enforcement, please email us at tips@buzzfeed.com. Or, to reach us securely, see this page.

An FBI spokesperson declined to comment on the agency’s investigative tools and techniques, though records seen by BuzzFeed News list individuals at more than 20 bureau offices as having run over 5,800 Clearview searches as of early 2020. Those same records show that employees at US Customs and Border Protection as having registered more than 270 accounts and run nearly 7,500 searches, the most of any federal agency that did not have a contract with Clearview at the time.

Asked how the agency uses the software in its policing work, a CBP spokesperson told BuzzFeed News it “does not use Clearview AI for its biometric entry-exit mission.” The spokesperson did not answer further questions.

Since BuzzFeed News first reported last year on ICE’s more than 8,000 searches through Clearview, the agency has expanded its relationship with the facial recognition provider. ICE has said publicly that it “does not routinely use facial recognition technology for civil immigration enforcement,” but it signed a $224,000 contract with Clearview last August.

Clearview’s strategy of handing out free trials meant that its facial recognition software was often used without any oversight. Officials at 34 public entities told BuzzFeed News that they had found police officers or other public servants using Clearview without approval.

Even more concerning, representatives at 69 law enforcement and taxpayer-funded entities initially denied to BuzzFeed News that their employees had used the software — but after further examination, discovered that they had.

Police officials in Chula Vista, California, for example, were adamant that their department did not use any facial recognition technology in its work. “Our officers can’t sign up for something like that on their own,” Eric Thunberg, a captain with the organization’s investigations division, told BuzzFeed News in November.

But after a more thorough search, Thunberg determined that they had. A “small number” of officers signed up for a free trial in 2019 and used the software to investigate threats — like “a photo of a kid holding a gun or weapon” — against dozens of schools in their jurisdiction, he said. “Absent of your inquiry, we never would have known about it.”

Similarly, the Tacoma Police Department in Washington initially denied using Clearview before discovering that an officer “in an investigative capacity” ran nearly 800 searches during a free trial that lasted from November 2019 to November 2020. Spokesperson Wendy Haddow noted that “the officer said there were no arrests made that he is aware of from the searches.”

The Los Angeles Police Department, the nation’s third-largest police agency, banned the use of commercial facial recognition in November following an inquiry from BuzzFeed News about the nearly 475 Clearview searches that officers ran as of early last year. John McMahon, the deputy police chief who oversees the department’s IT division, confirmed that the more than 25 officers and investigators who used the software did so in ways that were “not authorized.” The department declined to answer further questions.

In Alameda, California, BuzzFeed News found that police officers continued to use Clearview after elected leaders voted to ban the technology in December 2019. In the months before that vote, an Alameda police officer warned Clearview cofounder Richard Schwartz that the department was facing “an uphill battle” to approve a paid contact for the software. Schwartz’s reply, in an August 2019 email obtained via a public records request, denounced the “anti-facial-recognition narrative” and touted Clearview as a “state-of-the-art investigative tool for law enforcement that is super-accurate and 100% unbiased.”

He added, “Are they really going to let politics and deliberately misleading reports prevent you from using a life-saving tool like Clearview?” (The company declined to answer questions about Schwartz’s communications with Alameda police officers.)

When Alameda became the fourth city in California to ban the use of facial recognition in December 2019, some officers apparently did not heed the directive. Records seen by BuzzFeed News show that Alameda police officers — who ran nearly 550 searches in data — continued to use Clearview at least until February 2020, unbeknownst to city officials. The city manager and city council members, who told BuzzFeed News that the officers’ use of the software has been hard to track because of its free trials, are now investigating the matter.

“Never in my job would I ever think, Oh, I wonder if I can use this and not check it with a higher authority,” John Knox White, a member of the Alameda City Council, told BuzzFeed News. “If something is controversial, we should check in with the city attorney’s office; key decision-makers should be involved to make sure there’s no problem.”

He added, “That we have emails showing police used this technology after an actual vote saying you can’t do this is extremely troubling.”

Ton-That is an outspoken advocate for Clearview, parroting the company’s marketing claims in print and TV interviews. In February 2020, he said on CNN that Clearview was “99% accurate” and compared its achievements to the breaking of the sound barrier. “It’s gotten to the point where it works better than the human eye,” he said of the facial recognition software.

But these claims have not been vetted by an independent third party, and Clearview offers no research to support them. The company’s marketing materials claim that its recognition software is 98% accurate per 1 million faces, according to the benchmark created by the University of Washington’s MegaFace dataset; however, this has never been independently verified by the school. Ton-That declined to provide evidence that the technology had been reviewed by a third party but said that the company “plans to participate” in tests “to further validate the accuracy and reliability of Clearview AI.”

The University of Washington did not respond to requests for comment. Meanwhile, representatives from multiple law enforcement agencies told BuzzFeed News that they had opted not to purchase Clearview subscriptions because the software either did not work as well as they’d expected or did not meet department standards.

Dennis Natale, a detective with the Melrose Park Police Department in Illinois, which ran more than 120 searches as of early 2020, said that Clearview “didn’t directly affect” any of his cases. Lt. Jon Bowers of the Fort Wayne Police Department in Indiana told BuzzFeed News he didn’t recall any instance in which Clearview “was a lynchpin” in closing a case; he noted that his department pays for the software and uses it infrequently because it rarely returns matches on searches.

Shawn Vaughn, public information officer for the police department in Texarkana, Texas, which had run more than 280 Clearview searches as of February 2020, said the agency stopped using the software after a few months because of concerns “about its reliability and privacy issues.”

“As far as any claims of 100% accuracy by a product, it’s been my experience that those should immediately be viewed as suspect,” he said.

Part of the problem, officers explained, is that Clearview doesn’t seem to work well with the low-resolution photos and grainy surveillance footage that are so common in police work. Bowers said that Fort Wayne police officers used Clearview to identify alleged rioters during a Black Lives Matter protest in late May 2020, but only because the event was widely documented in pictures and video — something that is not true of the typical crime scene.

“The quality is often not good enough to put it in Clearview and find the [suspect in the] armed robbery from Friday night,” he said, noting that the department was reconsidering its $4,000 yearlong contract with the company.

Historically, concerns about the accuracy of facial recognition systems have been particularly acute when it comes to scrutinizing the faces of nonwhite people. Technology developed by other companies has been plagued by claims of racial bias and false identifications, leading to innocent people being accused of crimes. In 2018, the ACLU reported that Amazon’s facial recognition system incorrectly matched 28 members of Congress to people who had been arrested for crimes. There are now at least three known instances of people being jailed after being falsely identified by other facial recognition tools. All three were Black men.

Last June, in light of racial justice protests and further scrutiny on potential bias in facial recognition, Microsoft placed a temporary moratorium on the sale of its technology to police departments “until there is a strong national law grounded in human rights.” Amazon did as well for a year, saying it wanted to “give Congress enough time to implement appropriate rules.”

“The technology is not built to identify Black people,” Mutale Nkonde, CEO of the nonprofit communications group AI for the People, told BuzzFeed News. She noted that when facial recognition tools misidentify people of color, it amplifies dangers they already face from implicit racial bias.

Ton-That has repeatedly said that Clearview is immune to such errors. In a statement to BuzzFeed News, he reiterated this claim and said that “As a person of mixed race, ensuring that Clearview AI is non-biased is of great importance to me.”

He added, “Based on independent testing and the fact that there have been no reported wrongful arrests related to the use of Clearview AI, we are meeting that standard.”

However, Clearview did not provide any information about that testing, despite repeated inquiries. Furthermore, while BuzzFeed News found no evidence of wrongful arrests, we learned of false positives, suggesting the software is susceptible to similar problems.

Detective Adam Stasinopoulos of the Forest Park Police Department in Illinois, which stopped using Clearview after its free trial expired, said he saw false positives in search results within the facial recognition app. “I know that there were matches that weren’t exactly accurate,” Stasinopoulos said. BuzzFeed News reviewed documents in a Forest Park arrest in which Clearview correctly identified a suspect, a Black woman, in the theft of an engagement ring, yet it also matched her image with three other Black individuals who were not involved in the crime.

In early 2020, a source with access to the Clearview AI mobile app conducted a series of searches on behalf of BuzzFeed News. That person, who asked not to be identified for fear of retribution, ran more than 30 searches using a group of images including several photos of computer-generated faces.

On two searches for computer-generated faces — a woman of color and a girl of color — Clearview returned images of real people as matches. For example, it matched an artificial face to an image of a real girl of color in Brazil, whose school had posted her picture to Instagram. For searches of computer-generated white faces, no false matches occurred.

Clearview declined to respond to questions about these issues.

In Clearview contracts seen by BuzzFeed News, the company warns paying agencies that matches should not be the sole basis for making an arrest. But experts worry that the company’s unproven claims of accuracy could encourage investigators to use a possible match and develop cases they wouldn’t otherwise pursue. Facial recognition software is so new that lawmakers and the public are only beginning to grapple with how it is used.

Sen. Chris Coons, a Delaware Democrat and the chair of the Senate Judiciary Subcommittee on Privacy, Technology, and the Law, told BuzzFeed News that as facial recognition becomes more widespread, local and federal authorities need to enforce restrictions.

“As this reporting shows, we have little understanding of who is using this technology or when it is being deployed,” he said. “I’m also deeply concerned by evidence that facial recognition technologies can be error-prone or demonstrate harmful biases, especially when used on people of color. Without transparency and appropriate guardrails, this technology can pose a real threat to civil rights and civil liberties.”

Police can easily abuse and weaponize facial recognition tools like Clearview’s, Joy Buolamwini, founder of the Algorithmic Justice League, told BuzzFeed News, regardless of whether the tool is accurate.

“We have to take the conversation beyond questions of accuracy or algorithmic bias and start focusing on questions of algorithmic harms and civil rights,” she said. “When these systems are used as tools of mass surveillance, the excoded, those who are harmed by algorithmic systems, suffer the brunt of overpolicing, undue scrutiny, and false arrests.”

And then there is the issue of transparency. Michael Gottlieb, a lawyer for a racial justice protester who was arrested in Miami in May, learned Clearview had been used to identify his client as a suspect only after a local NBC affiliate station reported it. The Miami Police Department did not return multiple requests for comment.

Similarly, for those who were arrested during Black Lives Matter protests in Fort Wayne, Indiana, the police department there declined to disclose who had been identified with the assistance of Clearview.

Emails between police officers from late 2019 obtained through a public records request show some discussing whether to omit mention of Clearview AI in official reports.

“When you write reports about identifying suspects through Clearview AI,” one officer from Springfield, Illinois, asked a peer in Gainesville, Florida, in November 2019, “do you name the program, etc.? Or is it better to keep it vague. I wasn’t sure if there were any issues with trying to keep the information away from suspects about how we got their information.”

The Gainesville officer replied: “You can keep it general and say ‘through the use of investigative techniques. It’s not going to be totally wrong if you mention it.”

The rapid development and proliferation of commercial facial recognition has left lawmakers across the country scrambling to define digital privacy and protect our faceprints. It’s also made some police departments uneasy over deploying such technology in their communities.

Sheriff Rob Streck in Montgomery County, Ohio, didn’t mind that two of his detectives had tested Clearview on a trial basis, but he opted not to adopt the tool officially.

“I think there are still many unanswered questions about privacy and the technology,” he said.

Law enforcement leaders in other states echoed Streck’s concerns. The Utah Department of Public Safety considered using Clearview AI but decided against it because it couldn’t verify how, and from where, the company was mining its faceprints, Chief Brian Redd told BuzzFeed News.

“If a company is not being transparent or not wanting to give us information as to how they are mining their data, that’s a red flag for us,” he said. “Our community also had concerns about the use of their public photos, so we shut it down.”

After testing Clearview, the Utah Department of Public Safety continued to work with state leaders on a facial recognition bill, which passed through the state legislature earlier last month. It’s not as strong as some other states’, but it does require officers to submit a written request before running a search, and only allows them to do so in a felony case or a life-or-death situation. The bill also bans state agencies from accepting free trials from private companies like Clearview.

“This is a powerful tool, and it needs accountability,” Redd said. “When it comes down to it, we didn’t want to be the only ones who have the power to decide how to use it.”

Jameson Spivack, an associate at Georgetown Law’s Center on Privacy & Technology, said that while such guardrail policies are a good first step, they’re not worth much if they are not enforced properly.

“If you were to ask law enforcement how they use face recognition, they could just say, ‘Oh, it’s just an investigative lead,’ which may or may not be true,” Spivack said. “The fact of the matter is, in most cases, there’s nothing holding them to this. There’s, in most cases, no laws, in most cases, no internal policies. They pretty much have free rein to use it how they want.” ●

If you have information about Clearview AI, or other facial recognition technology used by law enforcement, please email us at tips@buzzfeed.com. Or, to reach us securely, see this page.