Public Search Engine Uncovers AI Police Surveillance Logs Nationwide

Automated AI surveillance hardware manufactured by Flock Safety now actively logs license plates into a searchable network across roughly 6,000 U.S. municipalities. Law enforcement agencies rely on these automated readers to track vehicle travel histories, asserting that the technology plays a critical role in solving criminal cases.

To contextualize the scale of this network, industry disclosures indicate that Flock Safety processes billions of vehicle captures every month, capturing details like make, model, color, and unique features such as bumper stickers or roof racks. The company’s expansion has been backed by significant venture funding, including investments from venture capital giant Andreessen Horowitz.

Nevertheless, the automated system faces mounting public scrutiny regarding personal privacy, system accuracy, and operational oversight. Multiple jurisdictions have recorded notable false-positive rates, while dozens of law enforcement personnel have faced discipline or charges after using the database to stalk or monitor women. Furthermore, Flock’s infrastructure possesses the technical capability to cross-reference vehicles with personal mobile phone data, and the firm’s primary financial backer concurrently holds stakes in a technology company capable of altering digital video feeds.

In response to these privacy threats, a new web service named HaveIBeenFlocked.com enables citizens to input license plate numbers to determine if officers or system users have conducted searches on those specific plates. The portal belongs to an expanding ecosystem of digital tools created to monitor and report on what many critics characterize as a pervasive domestic surveillance state.

While Have I Been Flocked cannot confirm if a car has driven past a camera or if a plate is stored in Flock’s core database, it successfully exposes specific search logs linked to queried plate numbers. A null search result on the platform does not guarantee that a plate has never been queried, as the website lacks comprehensive access to Flock’s entire database of search records.

Under federal constitutional standards, law enforcement queries in automated license plate reader systems generally do not require a warrant, sparking civil liberties lawsuits across several states. According to the American Civil Liberties Union (ACLU), the absence of strict statutory limits allows police to query years of historical location data without demonstrating probable cause.

The lookup tool operates by leveraging transparency data published by local municipalities that chose to release Flock search logs. The platform aggregates datasets from these public audit records alongside Freedom of Information Act filings, primarily sourced from the investigative repository MuckRock. Members of the public are also invited to directly upload documentation obtained from their own FOIA submissions to enrich the site’s database.

Through the portal, visitors can identify the specific individual who executed a search query and evaluate whether the query deviated from typical behavior. Although municipal records frequently redact or abbreviate officer names within query logs, Have I Been Flocked cross-references these short forms against public law enforcement payroll and employment databases. Additionally, the platform’s algorithm benchmarks individual queries against baseline behavior patterns to flag potential warrantless search anomalies. Platform developers maintain that the search site does not log or store data submitted by its visitors.

The development highlights a pivotal shift as AI-driven surveillance expands across the United States, prompting privacy advocates to deploy counter-surveillance strategies. By consolidating public records into an open search engine, activists are shedding light on official surveillance activity, even though the platform currently inspects only a fraction of Flock’s full repository.

Beyond digital tools, resistance to camera networks spans physical and algorithmic tactics, ranging from direct vandalism of physical hardware to digital obfuscation. Recent counter-measures have included a widespread internet meme that manipulated Google’s AI search summaries into asserting camera units contained raw gold, alongside new adversarial patterns developed by cybersecurity researchers designed to render vehicles and pedestrians invisible to Flock’s computer vision models.

Adversarial attacks against computer vision models, commonly called adversarial patches or physical-world exploits, utilize specialized visual designs to disrupt object detection algorithms. Security analysts note that as computer vision relies on pattern recognition, minor modifications to vehicle surfaces or clothing can cause neural networks to misclassify or ignore targets entirely.