Clearview AI Tests InquiryIQ Tool for Automated Web Profiling and Police Intelligence

Facial recognition developer Clearview AI has engineered and internally tested an unreleased artificial intelligence feature designed to automatically gather personal data across the internet and construct detailed investigative profiles on individuals. Discovered through source code embedded in public login files on the company website, the experimental software system, designated InquiryIQ, expands Clearview beyond face-matching searches by autonomously gathering information regarding suspect associates, online aliases, home addresses, current employers, physical attributes, and criminal record entries.

The revelation comes as Clearview expands its underlying facial image database past 70 billion harvested images while maintaining commercial access exclusively for vetted law enforcement and federal government entities under a 2022 legal settlement with the American Civil Liberties Union. Although Clearview maintains that InquiryIQ remains an unreleased prototype that has never been pitched or deployed to external law enforcement agencies, the discovery highlights how artificial intelligence developers are moving toward automated open-source intelligence gathering.

According to user interface code and technical documentation reviewed by journalists, InquiryIQ is structured as an automated assistant designed to streamline the labor-intensive research detectives perform after securing an initial facial match. When an investigator identifies a primary subject, the system initiates web and image searches, navigates online domains, and applies automated facial recognition to new photographs encountered across public websites.

The software then aggregates these disparate data streams into a structured network diagram labeled a Candidate Graph within internal files. This system systematically populates individual profile fields with telephone numbers, historical addresses, suspected social media accounts, known acquaintances, and employer histories. Clearview Chief Executive Officer Amos Kyler stated that the prototype was created to evaluate hypothetical investigative leads rather than act as an autonomous decision-maker, explaining that the software tests whether a specific piece of information can be validated or disproved.

Code analysis reveals that InquiryIQ incorporates demographic fields including age, gender, and race directly into its user interface, instructing the software that such parameters assist the system in making smarter search decisions. To drive these analytical processes, Clearview benchmarked several external large language models during testing, including software from SpaceXAI, the entity operating the Grok language model following the merger of SpaceX and xAI.

The interface also contained model selection toggles for Amazon Bedrock, an enterprise service facilitating access to third-party artificial intelligence frameworks. Addressing the presence of these system options, Kyler clarified that the model controls were deployed solely for internal benchmarking by company engineers rather than user customization by police officers. "No law enforcement user has ever used it, period," Kyler emphasized, adding that Clearview continuously evaluates external artificial intelligence architectures without having formally designated any model for public release. In a statement addressing the tool, Amazon indicated that AWS was not involved in InquiryIQ development and noted its terms do not prohibit law enforcement usage of Amazon Bedrock.

Privacy scholars and legal analysts caution that automating digital surveillance fundamentally alters established constitutional protections by stripping away the practical resource limits that historically constrained police investigations. Andrew Guthrie Ferguson, a law professor at George Washington University who studies legal technology, characterized the automated aggregation of online footprints as digital rummaging. Ferguson noted that while traditional investigations required physical labor to connect disparate public records, automated tools eliminate those procedural barriers.

Woodrow Hartzog, a privacy researcher at Boston University, argued that existing statutory protections were designed under the assumption that manual friction would naturally limit government surveillance capacities. Hartzog expressed skepticism regarding safeguards requiring human officers to review automated results, stating that "a human in the loop is a little bit of a cold comfort" as investigators routinely come to rely on machine-generated output. Court filings in Minnesota federal court, such as United States v. Sant, demonstrate how law enforcement officers have previously accepted misidentified Clearview search outputs during criminal investigations without conducting independent secondary verification.

The potential reliance on generative artificial intelligence for criminal investigations introduces significant evidentiary challenges regarding algorithmic hallucination and search auditability. Defense attorneys point out that large language models trained on open web content can generate factual errors, biased conclusions, or fabricated online links that fail to meet legal standards for establishing probable cause.

Michael Price, litigation director at the National Association of Criminal Defense Lawyers Fourth Amendment Center, questioned the evidentiary validity of using probabilistic language models to justify police actions. "A hallucination-prone chatbot would not be trusted as an informant under any other regular circumstances," Price stated, highlighting that models can internalize unverified internet postings. However, Ferguson observed a potential administrative benefit, noting that if an automated tool maintains comprehensive records of its search queries, model parameters, and data sources, it could produce a more transparent audit trail for court disclosure than traditional, unrecorded police research.

Clearview's internal testing of InquiryIQ coincides with continued commercial adoption across state and federal government bodies despite ongoing regulatory scrutiny. Founded in 2017 with early venture backing from investor Peter Thiel, Clearview expanded its user base by providing law enforcement agencies with searchable access to billions of photos harvested without user consent from platforms including Facebook, YouTube, and Venmo.

Following widespread litigation and international regulatory bans across Europe and Australia, former Chief Executive Officer Hoan Ton-That departed his role in late 2024, with Kyler assuming leadership in October 2025 to focus on technical oversight and operational auditing. Despite past controversies, Clearview maintains contracts with more than 2,000 police departments and federal agencies, including a $225,000 one-year contract awarded by U.S. Customs and Border Protection in February 2026 for tactical targeting and counter-network analysis.

Clearview maintains that InquiryIQ remains an internal software prototype without a defined timeline or authorization for commercial deployment. Company executive leadership confirmed that no external agencies have been granted access to the automated profiling assistant, and any future deployment would require strict verification protocols where human analysts must independently certify data points before adding them to official case records.

As regulatory agencies and criminal defense attorneys monitor the evolution of law enforcement tools, the discovery of InquiryIQ underscores how rapidly generative software is being integrated into surveillance technology. Federal agencies continue to utilize Clearview's primary face-matching database while legal experts press for clearer statutory rules governing automated intelligence gathering in law enforcement operations.