Analysis
WIRED reported on the morning of September 10 that Clearview AI has been quietly testing a prototype called InquiryIQ that pairs its facial-recognition search with a model from Elon Musk's xAI, the company behind Grok. The prototype surfaced only after reporters found code buried in files that Clearview's login page sends to any visitor's browser before they even sign in -- not through any company announcement.
InquiryIQ can run web and image searches, browse pages, and trigger Clearview's own facial-recognition system on photos it encounters along the way -- pairing that search loop with a model from xAI, which Pulse has covered extensively since its merger with SpaceX. After a law-enforcement user runs an initial face match, the tool builds what Clearview internally calls a "Candidate Graph" -- a web of possible identities and associates, compiling addresses, phone numbers, employers, aliases, social accounts, and arrest histories into a single profile. Clearview told WIRED the tool remains an internal prototype that has never been pitched to customers or used by law enforcement.
“Clearview told WIRED the tool remains an internal prototype that has never been pitched to customers or used by law enforcement.”
Clearview, founded in 2017 by Hoan Ton-That and backed early on by Peter Thiel, built its business on scraping billions of images from the open web and social media -- a practice that has already drawn a BIPA lawsuit in Illinois, GDPR fines across Europe, and a 2022 ACLU settlement restricting most commercial sales of its database. Comparable identity-search products include PimEyes, a public reverse-face-search tool, and Palantir's law-enforcement data-fusion platforms, though neither combines a facial-recognition hit directly with an LLM-driven web crawl the way InquiryIQ does.
This wasn't a product launch Clearview chose to make public -- it was found in client-side code, the same way security researchers routinely find unreleased features baked into web apps before companies are ready to talk about them. That gap, between what a company says and what its own code reveals it has already built, is the real story here independent of whether InquiryIQ ever ships to a customer.
What the discovery leaves unresolved: Clearview says InquiryIQ has never been used by law enforcement, and there's no evidence yet that it has -- critics of the company would note that its own past assurances haven't always held up once a product actually shipped. The regulatory risk exists regardless of deployment status. Illinois's BIPA already restricts biometric data collection without consent, and the EU AI Act bars real-time biometric surveillance in public spaces with narrow exceptions. A tool that auto-generates dossiers from a face match, even in prototype form and even if never deployed, is precisely the capability privacy regulators on both sides of the Atlantic have been trying to get ahead of before it reaches deployment, not after.
