Clearview AI Explores AI-Powered Investigative Tools as Internal Prototypes Surface Amid Privacy Concerns

The facial recognition firm Clearview AI, which gained notoriety in 2020 for scraping billions of images from social media platforms to populate a massive database for law enforcement, is currently exploring a new frontier in digital surveillance. Recent technical analysis of the company’s web-based assets has uncovered an experimental tool dubbed InquiryIQ, a system designed to automate the labor-intensive process of online investigative research. While Clearview maintains that this tool is a mere prototype that has never reached the hands of police, its existence raises significant questions about the intersection of generative artificial intelligence and the expansion of state surveillance powers.

A Timeline of Digital Surveillance Evolution

The history of Clearview AI is marked by a rapid escalation from a niche software developer to a central pillar of American policing infrastructure. Founded in 2017 with seed funding from figures such as Peter Thiel, the company operated in relative obscurity for several years. By 2020, its business model—harvesting personal images from platforms like Facebook, Venmo, and YouTube—was thrust into the public eye, triggering widespread condemnation from privacy advocates and legal action from several tech giants.

Despite these hurdles, the company’s reach has grown exponentially. In 2020, Clearview’s database contained approximately 3 billion photos. As of 2025, that number has surged to over 70 billion. Today, the company claims its services are utilized by more than 2,000 law enforcement agencies nationwide. The transition in leadership from founder Hoan Ton-That to current CEO Amos Kyler, who took the helm in late 2024, signals a strategic shift. While Ton-That’s tenure was characterized by a "move fast and break things" approach that prioritized market dominance, Kyler has pivoted toward a rhetoric of refinement, emphasizing internal controls, auditing, and institutional integrity.

Inside the InquiryIQ Prototype

InquiryIQ represents a significant technical departure from standard facial recognition, which essentially matches a photograph to an identity. According to code fragments and interface documentation analyzed by WIRED, InquiryIQ is designed to act as an automated "analyst assistant." Once an initial identity is confirmed via Clearview’s primary search, the tool is intended to fan out across the internet to construct a comprehensive "Candidate Graph."

The tool’s functionality involves accessing web sources to extract details such as employment history, known associates, aliases, physical characteristics, and residence data. The interface indicates that the system prompts users to input demographic markers—including age, race, and gender—to refine the search logic.

Perhaps most striking is the tool’s integration with third-party generative AI models. The interface lists options to utilize models from xAI, the company behind the Grok chatbot, as well as Amazon Bedrock. This inclusion has drawn scrutiny due to the well-documented volatility of generative AI. Grok, in particular, has been embroiled in controversy for producing inflammatory, racist, and extremist outputs. While Clearview’s management asserts that these models were merely being tested for performance comparison rather than deployment, the potential for an automated system to be "fed" by hallucination-prone AI creates a significant risk of algorithmic bias and data contamination.

The Problem of Digital Rummaging

The development of InquiryIQ highlights a broader trend in criminal justice technology: the automation of "digital rummaging." Legal scholars and privacy experts argue that the traditional investigative process—which requires manual, time-consuming research—has historically acted as a practical check on the breadth of state power.

"The privacy protections we have in place right now were mainly built in a world that assumed a certain amount of friction in the ability of governments to collect information," explains Woodrow Hartzog, a privacy scholar at Boston University. By collapsing the time required for deep background research from days or weeks into mere minutes, InquiryIQ could facilitate a shift where "fishing expeditions" become economically and logistically feasible. When the cost of investigating a person drops to near zero, the criteria for initiating an investigation may similarly erode, leading to more frequent, less targeted surveillance.

Furthermore, the "human in the loop" defense—the argument that an officer must verify the AI’s findings—is viewed by many as insufficient. As seen in the case of United States v. Sant, where investigators were found to have accepted flawed matches in a report, the sheer volume of data produced by automated systems can lead to "automation bias," where human operators become mere rubber stamps for machine-generated results.

Industry and Regulatory Perspectives

Clearview AI has firmly rejected the characterization of InquiryIQ as a product nearing release. In an official statement, CEO Amos Kyler emphasized that no law enforcement user has ever accessed the tool. "The mission today is the same, but our focus is on ensuring that the product hits the kind of expectation of integrity that our customers expect," Kyler stated. He argued that the interface’s references to various AI models were strictly for the benefit of internal engineers, not for external end-users.

Amazon, whose Bedrock platform was identified in the interface, distanced itself from the project. A spokesperson for AWS noted that while they do not have service-specific policies prohibiting law enforcement usage, the company is not involved in the development of InquiryIQ, and users are expected to adhere to established responsible AI policies.

The broader implications of these tools are being closely monitored by legal organizations. Michael Price, litigation director for the National Association of Criminal Defense Lawyers’ Fourth Amendment Center, warned that integrating generative AI into probable cause determinations is a precarious legal strategy. "A hallucination-prone chatbot would not be trusted as an informant under any other regular circumstances," Price noted, highlighting the danger of using unverified AI outputs to establish the basis for warrants or arrests.

The Search for a Silver Lining

Despite the substantial privacy and reliability concerns, some experts see a narrow path for the constructive use of such technology. Andrew Guthrie Ferguson, a professor at George Washington University who studies AI in the legal system, suggests that if properly implemented, these tools could actually increase transparency in police work.

Traditional detective work is often opaque; investigators may discard leads or pursue hunches without leaving a traceable record. If an AI system like InquiryIQ were designed to log every prompt, search path, and model selection, it could potentially provide a more auditable trail than the current manual system. "It’s a silver lining in an otherwise potentially fraught and disruptive change," Ferguson noted.

However, this optimism is tempered by the historical pattern of law enforcement technology adoption. As Ferguson observed, the pattern of deploying advanced surveillance tech before the establishment of adequate legal frameworks is a recurring theme in American history. As Clearview and other firms continue to innovate, the debate over InquiryIQ serves as a microcosm for a larger, unresolved question: how to balance the speed and efficiency of modern artificial intelligence with the fundamental privacy rights of the citizens it is designed to monitor.

Future Outlook and Governance

As of mid-2025, the trajectory of InquiryIQ remains uncertain. Clearview’s management insists that their current directive is to rapidly prototype new capabilities given the lowered barriers to entry in AI development, but they have provided no timeline for a potential product launch.

The pressure on the company is unlikely to subside. With the increasing number of lawsuits and the ongoing scrutiny from international regulatory bodies regarding the company’s massive biometric database, the introduction of a generative AI assistant is guaranteed to draw further attention from legislators. Whether this tool becomes a standard feature of modern policing or a cautionary tale about the limits of AI in law enforcement will likely depend on the willingness of judicial bodies to impose strict evidentiary standards on the "black box" of machine-generated intelligence. For now, the tool exists in the digital shadows—a prototype that illustrates both the power of current technology and the profound risks of its unchecked application.