Clearview AI is currently testing an artificial intelligence tool designed for automatically gathering information about individuals following facial recognition. This prototype, named InquiryIQ, can link data from the open internet, including social media accounts, addresses, employers, and potential connections, according to WIRED.
Journalists discovered this tool within files transmitted by the Clearview AI login page to browsers prior to user login. These files contained interface code, instructions, warnings, and descriptions of features. WIRED was unable to access the server-side of InquiryIQ, preventing independent verification of its claimed capabilities.
From Facial Recognition to Digital Profiles
Historically, Clearview AI has positioned facial recognition as a method to provide leads for further investigations. If a match was found, personnel had to manually verify any associated information about the individual.
InquiryIQ automates a portion of this subsequent process: after receiving initial data, the system autonomously searches the internet and images, opens pages, and analyzes the photographs it finds.
The prototype generates a Candidate Graph based on its findings, outlining possible identities and connections. The profile may include alleged addresses, phone numbers, employers, aliases, social media pages, arrest records, and associated individuals.
The interface also allows for the specification of age, gender, and race. According to the code reviewed by WIRED, these parameters are intended to assist the AI in making "more informed decisions" during searches. However, the exact impact of demographic characteristics on outcomes remains unclear.
The company asserts that law enforcement has never utilized InquiryIQ. The prototype has not been offered to clients and will not be released in its current form.
Clearview AI cautions within the interface that automatically obtained demographic information, social media accounts, and arrest records may be inaccurate. Users must accept or reject each finding and confirm that they have independently verified it. However, Boston University law professor Woodrow Hartzog expressed skepticism about whether this is sufficient:
“Human oversight is a weak consolation,” he noted.
He believes that over time, personnel verifying the outputs of the automated system may become merely formal validators of its decisions.
Different AI Models and Automation
The InquiryIQ interface offers a selection of systems for analysis. Among the options, WIRED identified models from SpaceXAI and Amazon Bedrock—a platform that provides access to third-party development tools. Amazon representatives commented to the publication that AWS is not involved in the development of this tool.
Experts consulted by WIRED indicated that such systems could significantly reduce the manual labor that previously took investigators hours or even days.
Hartzog pointed out the downside of automation. The labor-intensive nature of gathering disparate information previously limited the scale of surveillance, as checking large numbers of individuals was costly and cumbersome. InquiryIQ potentially lowers this barrier by automatically linking information that once required manual searches across various sources.
Andrew Guthrie Ferguson, a professor at George Washington University, highlighted a possible advantage of such systems. If they retain prompts, search queries, model selections, and procedural sequences, the investigative process could become easier to audit later.
“This is the bright side of a potentially problematic and disruptive change,” Ferguson remarked.
Clearview AI claims its database currently holds over 70 billion images from publicly available internet sources. According to the company, more than 2000 U.S. law enforcement agencies utilize this technology.
In 2023, reports emerged that American police had used the Clearview AI facial recognition system nearly 1 million times. Company CEO Hoan Ton-That stated that hundreds of law enforcement officers across the country use the startup's technology, estimating the database at around 30 billion images.
Previously, French authorities fined Clearview AI €20 million for unlawfully processing personal data and demanded that it cease collecting information about the country's citizens.