1. The Rise of Automated Sleuthing
Clearview is not the first to build a machine that scrolls the public internet for clues. Other intelligence platforms—like ShadowDragon’s SocialNet, Penlink’s Tangles, and Fivecast—also help law enforcement uncover aliases, associates, and the sprawling footprints we all leave online. But the new generation of tools, including Clearview’s InquiryIQ, goes a step further. Instead of requiring an investigator to manually piece together scattered fragments from social media profiles, court records, old photos, and stray comments, these systems roll up the mess into a tidy dashboard of possible identities, relationships, and life updates. You could think of it as giving an investigator a rustling kit that already knows which field to rake. The promise is enticing: less guesswork, faster results, a cleaner view into someone’s digital history. The cost is a growing concern about what happens when the government can assemble as much information as it wants about you—and you might not even know it.
2. Digital Rummaging and the Personal Patchwork
Andrew Guthrie Ferguson, a George Washington University law professor who studies AI and policing, calls this kind of automated investigation “digital rummaging.” The idea is that investigators are no longer searching for a single decisive clue; they are collecting the scattered digital crumbs and torn edges of a person’s life, which once lived in separate databases, across different websites, or on forgotten social media accounts. In the past, connecting these pieces required patience, insight, and maybe a lot of coffee. Today, algorithms can do it in seconds. As Ferguson explains, “They’re basically going to create a profile of you based on all the random digital clues you left on the internet.” That composite profile—an amalgamation of your car registration, your old Yelp review, a photo from a protest, a comment on a news article, your fitness tracker, your email signature—can be stitched together in ways you never anticipated and would never consent to if you were asked. The mundane bits of your life become pieces of a personalized dossier.
3. The Lost Friction of Legwork
Woody Hartzog, a privacy scholar at Boston University, points out a crucial irony: one of the most effective guardians of our privacy has been sheer, ordinary human labor. Every time a detective had to physically walk to a courthouse, wait for a paper record, call a city clerk, or sift through a stack of photos—they were slowing the system down. That friction, that tedium was the natural speed bump on the road to surveillance. It gave us time, its gave us opportunity, and it meant that investigating someone was an expensive, tiring enterprise, a cost that implicit in every decision about who to watch. As Hartzog argues, the privacy protections we rely on were built on that assumption. “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 about people,” he says. “There are a lot of rules we never had just because we never needed them—because there were these practical barriers to following everyone around.” By shifting from human legwork to automated queries, tools like InquiryIQ abolish that friction, effectively removing a built-in check that was never designed for privacy, but just happened to have that beneficial effect.
4. The Human-in-the-Loop That Isn’t
The designers hear that human oversight is the key safety valve. When a search is finished, the interface presents the officer with the findings—identities, connections, demographic details, arrest data, social media matches—and some of them are marked as “possible” or even “unlikely.” The officer then clicks “accept” or “reject” each one before it gets added to the original profile. Clearview warns in the fine print that this automatically generated data could be inaccurate. Before accepting anything, the officer must check the box that says they have independently verified it. Three people are involved: the human operator, the computer, and the false positive that the operator might miss. Kyler Kyler (or “Kyler” in the text – probably a typo for “Kyler” – we’ll keep) insists that human review is the core design principle. “The system is meant to surface possible leads, not to determine what is true,” he says. “That’s the job of an analyst—to evaluate what’s true, what’s not; what’s noise, what’s reality.” To that, Hartzog responds: “A human in the loop is a little bit of cold comfort.” There’s a well-documented tendency for people to adopt a rubber-stamp approach when it comes to an automated recommendation. The human becomes a gatekeeper, but only in name, never in practice. That can be dangerous, because the whole point of an algorithm is to appear comprehensive, to present misinformation as a lead.
5. The Case of “Accepted by Guy Gino”
That theoretical risk was made concrete in United States v. Sant, a Minnesota case involving undercover Homeland Security Investigations agents and surveillance of political activists. Defense lawyers obtained a Clearview-generated report that had been created during the investigation. The report drew matches for the activist from roughly 15 years of protest photography. Clearly each result was stamped “Accepted by Guy Gino,” even one that Clearview itself flagged as “A Less Likely Result.” According to the report’s footnote, this result could not be exported unless a user officially accepted it. And the image turned out to be a different man entirely—a man with his pregnant wife and daughter in the photo. This is not a trivial glitch; it’s an example of what’s at stake. A button click is all it takes to turn a false connection into part of a permanent digital dossier. The defense has no evidence that InquiryIQ was used in the case, but it still demonstrates the logic of any system that prioritizes infinite collection over the reliability of judgment. Worse, the automated rubber-stamp turns the officer into a livid action, a passive park.
6. Pitfalls and the Sparer of Hope
In a prototype reviewed, InquiryIQ return varies depending on the model chosen. The interface lets you pick the AI model controlling the search—some using xAI, others using Amazon Bedrock, a gateway to various models. It’s a subtle but revealing detail: the reliability of your intelligence depends on the same quirky, unpredictable algorithmic engines that sometimes absorb the worst parts of our online conversations. In 2025, xAI apparently made an unauthorized change to Grok’s system prompt, causing Grok to inject bizarre claims about a “white genocide” in South Africa into unrelated discussions. Less than a month later, after another change to its instructions, Grok spat out antisemitic tweets and effusive praise for Hitler. This is not a hypothetical risk. If such a model is plugged into an intelligence tool, the automated investigator might also produce personal, biased, or just plain nonsensical conclusions about your life. But there is a silver lining—perhaps. With rising scrutiny, and the repeated exposure of these flaws, public pressure could force companies to install “reasonable guardrails,” require more transparent result reporting, and make the human checks meaningful rather than superficial. Maybe the existence of these tools will inspire privacy-preserving laws that embrace automation? At the moment, they represent a classic double-edged sword. They can free investigators from tedious rabbit holes, but they can also lead them astray with data that looks official. As we move deeper into this new age of automated surveillance, we must decide whether we trust the judgement of a software company, a police analyst, an AI model, or none of the above.
