In a move that places it at the forefront of biometric policing in Virginia, the Bristol Virginia Police Department has formally launched a facial recognition technology (FRT) program that is designed to bolster investigative capacity while enforcing strict accountability and civil liberty protections.
Bristol is a city with under 20,000 residents next to the Tennessee border. The police department’s adoption of facial recognition places it within a small cohort of law enforcement agencies across Virginia that are leveraging biometric tools within the boundaries of state law. However, the thoroughness of Bristol’s FRT policy, its publicly accessible documents, and its commitment to annual oversight, distinguish it from its counterparts. In a national climate where transparency and surveillance are often at odds, Bristol has charted a path that emphasizes structure, consent, and accountability.
The facial recognition algorithm that the Bristol Police Department is using – approved by the Virginia Division of Purchases and Supply – is that of Clearview AI, which the department accesses through what is believed to be a heretofore unknown partnership with LexisNexis Risk Solutions. According to chief of police D. Byron Ashbrook II, who spoke to Biometric Update, Clearview AI is accessed through a “LexisNexis Accurint account,” the contract for which is through LexisNexis Risk Solutions.
Neither Clearview nor LexisNexis responded to a request for comment.
Officers can submit a probe image, such as one obtained from a security camera or a witness and receive a ranked list of potential matches.
Virginia lifted its statewide ban on law enforcement use of facial recognition in April 2022, replacing it with one of the strictest regulatory frameworks in the United States. The new statute, enacted as Senate Bill 741, requires that any agency must implement a policy that meets or exceeds the law’s standards before deploying FRT technology.
The Virginia Division of Purchases and Supply may only approve algorithms that have undergone National Institute of Standards and Technology (NIST) testing and demonstrate at least 98 percent true positive accuracy, with minimal demographic bias across race, skin tone, ethnicity, or gender. As of August 2024, 15 facial recognition algorithms have been approved for use by Virginia law enforcement.
Effective June 1 and governed by a 27-page comprehensive directive – General Order 9-09 – the Bristol Police Department’s FRT program is notable not just for its technological capabilities, but for the structure of oversight, transparency, and legal compliance it has institutionalized at the municipal level. The department’s adoption of facial recognition reflects both a state-authorized initiative and a highly localized effort to address growing public concerns around biometric surveillance, algorithmic bias, and government overreach.
Under the leadership of Chief Ashbrook, the Bristol Police Department has framed the system as a public safety enhancement tool meant to expedite the identification of suspects, victims, and missing persons while minimizing misuse through layered procedural safeguards.
Ashbrook acknowledged the controversy surrounding Clearview AI, saying that he’d “read some of the stories/articles of [it] being used improperly as probable cause or credible source or used in a ‘live’ fashion, etc,” but assured that “that is not our intent, and, frankly, is counter to our policy and Virginia law. I wanted this solely in hopes to provide faster potential investigative leads for detectives and want to do so with restricted/controlled access.”
Ashbrook was clear that “if we don’t see value in the program, we will not continue using it.” He emphasized that Clearview AI “is only to be used as an investigative lead, not probable cause and should be treated like an anonymous tip, I can’t stress that enough.”
Crucially, the system is not connected to live video streams, real-time surveillance feeds, or body-worn cameras, and it cannot perform face recognition on video content. Every search is bound to an explicit, documented law enforcement purpose as outlined in the department policy’s 18 permissible use cases.
Unlike some police departments across the country that quietly implement facial recognition behind closed doors, the Bristol Police Department has opted for a more transparent route. The department publicly posted its facial recognition policy and integrated it into its formal governance processes.
The policy itself includes annual audits, semi-annual compliance reviews, training and certification standards, and the implementation of the Fair Information Practice Principles as a privacy foundation. This marks an intentional contrast with jurisdictions that have faced lawsuits or backlash for facial recognition misidentification.
“As to our training and usage, we are doing this in a very restrictive manner so as to prevent any misuse of the program,” Ashbrook told Biometric Update. “Only two employees of this agency have access to run images through facial recognition — the Criminal Investigation Division (CID) lieutenant, and in an emergency or if the CID lieutenant is not available, our deputy chief/major.”
Ashbrook said, “there are strict guidelines as to recordkeeping and justifications for utilization, so we are in compliance with” the Virginia Department of Criminal Justice Services.” Ashbrook noted that he and the two other officers authorized to use FRT “have been through Clearview AI’s two-hour training program” and have been provided “required reading in reference to cultural competency and FRT, etc., to ensure they have knowledge, and we meet policy.”
The chief added that “training … will be ongoing since it is such a new frontier,” and is “in the process of having [his] CID lieutenant set up for training with the FBI this fall.”
To operate the system, officers must undergo specialized training. Only those with advanced instruction are permitted to conduct searches, evaluate image quality, or interpret results. Probe images are evaluated against known image repositories, and candidate matches must be confirmed through manual morphological comparison by trained examiners. The department explicitly bars the use of artist sketches, AI-generated morphologies, or manipulated renderings for facial recognition searches, except in narrow cases such as age-progressed images of missing persons.
Furthermore, the resulting candidates are treated solely as investigative leads –never as definitive identifications. Accompanying each returned result is a disclaimer reminding the recipient that the image is not to be construed as a positive identification.
Mobile searches are subject to even tighter scrutiny. Officers are prohibited from conducting facial recognition queries via handheld devices such as smartphones. Only laptops and tablets issued by the department and capable of supporting high-resolution image comparisons may be used.
These devices must be connected to the secure police network, not public cellular internet, and must authenticate each user before access is granted. The login screen also reiterates usage restrictions and privacy requirements set by Clearview AI. This level of layered authentication and controlled access underscores the department’s stated commitment to limiting function creep and unauthorized access.
The policy’s drafters were particularly mindful of First Amendment protections. The use of facial recognition at protests, rallies, or other constitutionally protected gatherings is limited by clear language stating that the technology shall not be used to surveil lawful activities.
If the department deems it necessary to use facial recognition during such events — for example, in response to violence or an emergent threat — a justification must be documented in advance. This includes specifying the basis for collection, identifying the officers involved, and outlining how images will be handled and retained.
Oversight is centralized under the Bristol Police Department’s CID lieutenant, who also acts as the program administrator. This role includes approving or denying search requests, managing user credentials, auditing the system’s use, and ensuring data purging protocols are enforced. Search logs are maintained for up to two years for audit purposes but are not searchable and cannot be used to re-initiate identification efforts.
Anyone who accesses the system without authorization or in violation of policy faces disciplinary action and criminal penalties, including a Class 3 misdemeanor under Virginia state law.
Data sharing with outside agencies is allowed only under strict conditions. An external law enforcement agency must enter into a memorandum of understanding with the Bristol Police Department and undergo requisite training before submitting a search request. If a search is performed, the agency is informed that the results are leads — not confirmations — and cannot be used as evidence without further investigation.
Joint investigation results are also subject to disclosure rules that prohibit publishing, selling, or redistributing data unless required by law and authorized by the originating agency. The department reserves the right to deny search requests at its discretion.
Importantly, no images are stored in the facial recognition system unless they become part of an active investigation. Probe images are maintained in accordance with their case file designations and may be added to an unsolved image file for periodic review. Any image that becomes legally classified as evidence is reviewed by the CID lieutenant to determine whether extended retention is warranted.
The security framework surrounding the system is robust. Bristol police employ multi-factor authentication, encrypted communications, and role-based access control to prevent unauthorized use. Passwords must meet complexity requirements and are rotated every 60 days. Audit logs record every access point, including user identity, case number, search details, and the justification for each query. These logs are retained per guidance from the Library of Virginia and are subject to random semi-annual audits.
Bristol’s implementation does not stop at internal compliance. In accordance with Virginia law, the department submitted a notice of intent to the state’s Division of Supply and Purchases and received approval for its vendor selection. It has also entered into a memorandum of understanding with the Department of Criminal Justice Services through which it submits annual use data.
This data includes the number of queries performed, match rates, case closures linked to facial recognition leads, offense types, and demographic details related to images queried. The department is also obligated to maintain a detailed user query history to facilitate any future criminal proceedings or appeals.
The policy also includes a redress mechanism. Individuals harmed by an erroneous identification or who wish to challenge the accuracy of facial recognition data can file a complaint with the department. All complaints are investigated within 30 days and must be documented for auditing purposes. If a record is found to be inaccurate or improperly merged, it must be corrected or purged. Appeals processes are available for those unsatisfied with the department’s resolution.
Across Virginia, law enforcement agencies have experimented with or implemented FRT in recent years. Norfolk and Virginia Beach police both ran trials of Clearview AI before the 2021 ban, though public pushback and legal scrutiny prompted Norfolk to suspend its program.
Richmond and Chesterfield County police departments have included facial recognition tools within newly launched “real‑time crime centers,” combining it with license-plate readers and surveillance cameras. Although Richmond has not yet publicly detailed a complete policy like the Bristol Police Department, its deployment of FRT in conjunction with live data feeds contrasts sharply with Bristol’s static-only approach.
And here, the differences are stark. Bristol’s policy prohibits real-time tracking of individuals and uses encrypted, governed hardware for queries. In contrast, the real-time crime centers in Richmond and Chesterfield County hint at a broader adoption of surveillance technologies, including FRT, license-plate readers, and live camera feeds, raising heightened concerns among civil liberties advocates.
Critics across the state argue that while SB 741 represents a regulated framework, its latitude still permits potentially broad uses, with insufficient protection against misuse. Concerns linger over demographic bias in FRT algorithms, despite NIST-approved performance metrics, and questions remain about agencies’ transparency and oversight capabilities.
In Bristol, the strength of the safeguards — strict usage limits, audit trails, manual verification — offers a more cautious model compared to police departments that integrate FRT into active crime-centers. Whether this model becomes a standard for other small- to mid-sized police departments in the Commonwealth of Virginia remains to be seen. But if Bristol’s policy proves effective in both identifying threats and preserving civil liberties, it may well become a case study in how to deploy controversial technology in a way that earns public trust.
Article Topics
biometric identification | biometrics | Bristol Virginia Police Department | Clearview AI | facial recognition | law enforcement | LexisNexis | police | Virginia