Clearview AI Tests Tool for Police to Unearth Online Data

Clearview AI is currently testing a new artificial intelligence system designed to assist law enforcement in constructing detailed digital profiles of individuals. The company, which gained notoriety in 2020 for scraping over 3 billion images from the internet to identify suspects, is now exploring deeper data extraction methods. This experimental tool aims to go beyond simple facial recognition by uncovering personal connections, employment history, aliases, and physical characteristics associated with a specific name or face.

The unreleased software, known as InquiryIQ, functions as an automated research assistant. It takes initial clues gathered during a standard Clearview search and expands them across various web sources. By opening webpages and analyzing embedded images, the system attempts to compile a comprehensive dossier on the subject of an investigation. According to code analyzed by WIRED, the interface allows users to input demographic details such as age, gender, and race. These inputs are purportedly used to help the system make “smarter decisions” regarding its search parameters.

One of the underlying models tested for this functionality comes from SpaceXAI, the entity behind the Grok chatbot. Following a merger between SpaceX and xAI in February, this model has been integrated into Clearview’s testing environment. Grok has faced significant criticism for generating racist, extremist, and inflammatory content, positioning itself as an alternative to what its creator Elon Musk describes as overly cautious AI systems. It remains unclear how these demographic inputs influence the tool’s outputs, and Clearview declined to provide specifics on this mechanism.

The potential implications of such technology extend far beyond efficiency gains. Experts warn that tools like InquiryIQ could fundamentally alter police investigations by reducing weeks of manual detective work to mere minutes. While this speed might accelerate case resolutions, it also drastically lowers the barrier for conducting broad, low-cost inquiries into individuals who might otherwise escape scrutiny. Furthermore, because generative AI can yield different results from identical starting points, reconstructing the logic behind why a specific lead was pursued rather than another may prove nearly impossible.

Despite the sophisticated appearance of the prototype, Clearview maintains that InquiryIQ is not a finished product. CEO Amos Kyler states that the tool has never been pitched to or used by any law enforcement agency. He explains that the visible options for different AI models, including those from SpaceXAI, were included solely so engineers could compare performance metrics during internal testing. “No law enforcement user has ever used it, period,” Kyler asserts, emphasizing that the current iteration is strictly an internal experiment rather than a deployable service.

WIRED identified InquiryIQ through files automatically sent to visitors’ browsers before they log in to Clearview’s website. This method mirrors previous discoveries of other emerging AI tools, such as Flock Safety’s OS Investigate. The publicly accessible code reveals thousands of lines of text detailing the interface’s instructions, warnings, and features, providing insight into how the company envisions the tool operating. However, these files do not disclose the backend processes, the accuracy of the data collected, or the extent of internal usage.

Clearview argues that the polished nature of the prototype should not be misinterpreted as an imminent release. The company notes that modern AI development allows engineers to build complex interfaces rapidly, shifting the focus toward swift prototyping rather than immediate deployment. The directive given to Clearview’s team is to advance these prototypes quickly, leveraging new technological capabilities to explore potential functionalities without committing to a final product structure.