Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Surveillance Clearview Ai topic
No spam. Unsubscribe anytime.
Clear Lake Council hears police briefing on ClearView AI; vendor stresses "investigative leads" and training
Summary
Police demonstrated ClearView AI facial-recognition software and ClearView representatives said the tool provides investigative leads from a large open‑web image database; officials said searches are manual, accounts audited, and no passive camera feeds are being run into the system.
The Clear Lake City Council on June 6 heard a presentation from the police department and ClearView AI on a facial‑recognition product the department has been trialing for investigations.
Lieutenant Snyder told the council investigators had given "positive feedbacks" after the department ran a trial and that the department has added oversight measures: "The user will receive a mandatory training course ... Each user is required to complete an intake form with a case number and the type of an investigation for every search that they conduct. The user accounts are monitored and audited by the operations commander," he said.
Scott Geiser, representing ClearView AI remotely, said the company supplies investigative leads by matching a probe image against a large open‑web image corpus and returning ranked candidate images and source URLs. "We have over 40,000,000,000 images that have been gathered from the open Internet," Geiser said, and he described the tool as intended to speed investigations and provide leads that detectives then corroborate with traditional methods.
Both vendor and police emphasized limits on use. Geiser said the company does not ingest agencies' probe images into ClearView's dataset: "Those images that your department is probing across our database never become a part of our ClearView AI database," and Snyder confirmed searches are manual, not passive: "Absolutely no" passive scraping from ALPRs or surveillance is being done. Snyder said he currently audits accounts "every couple weeks" and that the department has five users on the platform.
Council members and members of the public asked about accuracy, demographic bias, juvenile images and safeguards. Geiser cited NIST testing and said the company's "wild category" performance is high; he added that a human must remain in the loop and recommended peer review steps for investigators. Council members pressed for regular audit schedules and limits on sharing: Snyder said records retention depends on the case and that investigations flagged as ongoing can remain live until closed.
Residents asked whether the system could help locate missing children or identify people from doorbell cameras; Snyder and Geiser said a clear image can generate leads but stressed the output is investigatory, not proof. No policy action or ordinance was proposed; the briefing was informational and the council took no vote on adopting a use policy at the meeting.
The council asked staff to continue monitoring usage, confirm audit cadence, and return if the department seeks a formal policy or an expansion of user access.
