Aurora, Colorado police move closer to using facial recognition technology

After years of internal planning and public debate about biometric surveillance, the Aurora Police Department (APD) has asked city leaders to formally allow it to use facial recognition software in criminal investigations, a move that has sparked questions about privacy, accuracy, and the proper limits of law enforcement technology.

The two systems are under consideration are ROC’s facial recognition service within LexisNexis’ Lumen/AVCC software platform, and Clearview AI. The department recently posted draft accountability reports for both systems to its website, opening a public comment period as required under Colorado law.

Under the draft proposal, Aurora would deploy facial recognition only in after-the-fact investigations when detectives already have reason to believe a crime has occurred and seek to identify persons of interest, witnesses, or victims.

The department emphasizes that algorithmic matches would not serve as probable cause for arrest. Instead, they would act as investigative leads requiring corroboration through traditional detective work.

To guard against misuse, Aurora’s draft policy outlines several layers of human review. Any candidate list produced by the software must be screened by an operator, subjected to peer review, and approved by a supervisor before a lead reaches detectives. The policy also prohibits live or continuous surveillance, immigration enforcement, harassment, and persistent tracking without a court order.

These provisions reflect Senate Bill 22-113, a 2022 Colorado law that created statewide guardrails for government use of facial recognition. The statute requires public notice and publication of accountability reports, formal human review of any outcome that could affect an individual, and a public comment process before deployment.

The law does not automatically authorize all forms of facial recognition. It prohibits real-time or continuous tracking unless justified by a warrant or court order, and mandates that agencies demonstrate necessity and oversight mechanisms.

In Aurora’s case, this means the city council must still approve any deployment, the department must host multiple public meetings, and the public comment period must remain open for at least 90 days.

While APD has discussed potential budget impacts with city staff, detailed cost projections have not been finalized publicly. Earlier internal estimates referenced a multi-year implementation plan and limited initial access for trained detectives, but those figures have not been confirmed in official city documents.

Aurora already uses LexisNexis’ Lumen system for information sharing with other Colorado law-enforcement agencies, which could allow integration without significant new infrastructure costs.

Even as city staff argue that facial recognition could improve investigative efficiency, opposition has surfaced over transparency, algorithmic bias, and the potential erosion of civil liberties.

Councilmember Alison Coombs, the lone dissenting voice when a committee advanced the proposal, warned that its “scope is extremely broad” and raised concerns about consent and access to personal data. Councilmember Curtis Gardner questioned the risk of wrongful identification, citing national cases where misidentifications led to innocent people being detained.

Supporters counter that state law and Aurora’s internal policy ensure multiple layers of human review and corroboration, but privacy advocates remain skeptical. The American Civil Liberties Union of Colorado’s Public Policy Director, Anaya Robinson, said facial recognition systems “have repeatedly misidentified women, people of color, and individuals with disabilities,” adding that “even when human review is required, the bias built into these systems can produce unjust outcomes.”

The two systems under review differ sharply in data scope. Lumen’s version, integrated with the Colorado Information Sharing Consortium, draws primarily from criminal-justice databases already accessible to law enforcement. Clearview AI, by contrast, scrapes billions of images from the open internet, raising deeper questions about consent, privacy, and the accuracy of matches generated from uncontrolled, real-world photographs.

Aurora’s draft accountability reports state that any use of facial recognition would be subject to internal audits and could be suspended if error rates or disparities exceed acceptable thresholds. However, the specific numerical benchmarks for such suspensions have not yet been published.

Colorado’s regulatory environment for biometric and AI technologies continues to evolve. In 2024, lawmakers passed House Bill 24-1130, a sweeping biometric privacy and AI accountability measure that took effect in July.

The law strengthens requirements for explicit consent, limits how public agencies and private entities can share or retain biometric identifiers, including facial templates, and mandates clear disclosures when automated decision systems are used in ways that affect individuals’ rights or access to services.

Together, SB 22-113 and HB 24-1130 place Colorado among the most tightly regulated states for facial recognition and biometric data, even as cities like Aurora weigh whether to authorize such technologies at the municipal level.

The proposal has cleared an initial city-council committee and is scheduled for formal readings before the full council, along with additional opportunities for public comment. If ultimately approved, Aurora would join a small number of Colorado jurisdictions that have explicitly authorized police use of facial recognition under the state’s regulatory framework.

Article Topics

biometric data  |  biometric matching  |  biometrics  |  Clearview AI  |  Colorado  |  criminal ID  |  facial recognition  |  police  |  ROC