Every role within law enforcement comes with its own set of challenges. If you’re on patrol, there’s an ever-present wonder about what dangers the next call could bring. If you’re an investigator working child exploitation cases, in addition to the emotional toll the job brings, you may often find yourself with few leads to work with, wondering how you’re going to solve the case.
As technology becomes more and more sophisticated, so too do the methods used by perpetrators who are involved in child exploitation. This has led to an exponential number of these types of crimes across the country, and when cases involve the welfare of a minor, every second matters when you’re working to uncover leads.
In many instances, leads are limited and investigators sometimes only have a video or still image of a child, making it difficult to determine who the perpetrator is, let alone who the victim is. Young children often don’t have the same online presence as adults, such as a driver’s license, bank account, or other type of identifying information, therefore investigators need to lean on other investigative tools to bring these minors to safety.
A TOOL WITH GUIDELINES
Used by federal, state and local law enforcement agencies across the United States, facial recognition technology (FRT) is becoming increasingly helpful in identifying the victims of child exploitation. Whether an investigator has a picture of a child or of someone they believe might be a perpetrator, tools like Clearview AI can help provide additional information leading to possible suspects.
“I see it as a tool and that’s all it is,” said Kevin Sibley, a 27-year federal law enforcement veteran and director of federal engagement at Clearview AI. “It’s not the end-all be-all – it’s not the solution to the problem. It is a tool to help identify the solution, but I see FRT as something that needs to be given limits, policies and regulations to control its usage.”
FRT is sometimes scrutinized by the public, particularly when it’s used for unclear purposes without much explanation as to how and why it is being used. Sibley stresses that people’s privacy concerns are legitimate, so it is important that when law enforcement agencies use FRT tools to promote public safety and justice for victims, they do so in a way that is sensitive to the protection of privacy and other constitutional rights.
Clearview AI has designed its FRT technology with these needs in mind. Clearview’s algorithm achieved very high accuracy scores across all tested demographics in the industry-standard Facial Recognition Technology Evaluation test by the National Institute of Standards and Technology.
HOW CLEARVIEW AI WORKS
It’s a common misconception that FRT looks at faces as a whole and searches for other faces as a confirmed match. With Clearview AI, the enhanced FRT breaks down an image into smaller pieces and then searches for potential matches.
“Like any modern facial recognition algorithm, Clearview AI’s algorithm uses complex mathematical formulas to generate a series of data points from the pixels in an image of your face,” explained Sibley. “It doesn’t characterize an individual by color, race or gender. It just sees points on a face and it creates an arbitrary numerical output – a long string of numbers.”
What is the tool comparing these long strings of numbers against? It uses them to compare against photos identified by its web crawlers which comb the internet, including news sites, social media sites and any other website that’s not behind a password-protected login, or a privacy setting enabled by a user of a website. It then gathers images and the URLs for those images and stores them in a database. Clearview AI’s database contains more than 40 billion images and that number will continue to grow in 2024.
“When you take Clearview’s robust facial recognition technology and you bounce it against a very large data set of open-source images from around the world, you very quickly come up with potential matches,” said Sibley.
He emphasizes that Clearview AI doesn’t identify “matches” explicitly, as investigators are expected to take additional steps to verify or refute a person’s identity. “We are a tip and a lead – we are not the last stage,” Sibley stresses.
Investigators using Clearview AI can stay organized by creating private folders for each case or can use the tool collaboratively by setting up shared folders. Supervisors within an agency can see the photo each investigator in their agency runs, and the technology records key statistics such as the number of logins, searches and results for each individual officer.
A SUCCESS STORY LIKE NONE OTHER
It might sound like a tall task to identify a potential image match among billions of photos from around the world, but Clearview AI’s success has been documented time and time again. Perhaps one of the most notable cases took place in Nevada in 2019. A web services provider was notified that one of its users received images that pointed to the sexual abuse of a young girl.
In one of the images, a male face was visible, but no other details were present to help investigators narrow down a list of potential suspects. With the help of Clearview AI, authorities found a different image online, one taken at a fitness tradeshow, that appeared to include the same man in the background.
Found on someone’s Instagram page, the photo included a specific Las Vegas location along with the name of a company that was present at the tradeshow. Investigators called the company, discerned who was working at the booth that day and, after finding additional confirming evidence, made an arrest.
“The perpetrator is an Argentine national and he wasn’t in any immigration database, DMV database or any other known U.S. government database,” said Sibley. “This was a four-month investigation and within a matter of minutes after using Clearview AI, they were able to get a solid lead to help identify who this individual was.”
The victim turned out to be a seven-year-old girl who was under the perpetrator’s supervisory control. The girl was rescued and the suspect was sentenced to 35 years in prison.
HELPFUL IN A RANGE OF SCENARIOS
Like in the success story above, Clearview AI is often able to help investigators uncover leads when all they have available is an image of an adult. However, the software can also be equally valuable when a child’s image is the only evidence present.
“It’s very good at helping to identify people who may have been posted to social media when they were a small child and now they’re a runaway teen, they’re being abused or they’re being forced into human trafficking or any type of child exploitation,” said Sibley.
Creative and experienced investigators can imagine many applications for Clearview AI’s facial recognition search engine. For example, it can be used for the purpose of force protection during covert operations in addition to providing investigatory leads.
“Through undercover operations, whether it’s child exploitation, human trafficking, drug transactions or money laundering transactions, officers are meeting with potentially unknown individuals,” Sibley explained. “They don’t know what this person’s criminal record is. They don’t know what their propensity for violence is. Clearview AI can be used in virtual real-time from a cover team taking a photograph of who they’re meeting with. From an officer safety standpoint in the undercover world, this is a very useful tool.”
Visit Clearview AI for more information.
