Flock Introduces Tool Aimed at Detecting Police Misconduct, but Specifics Are Hazy

Flock, a startup specializing in surveillance technology, introduced new policies and tools on Thursday designed to mitigate the misuse of its automated license plate camera systems and improve accountability for law enforcement users.

The company now recommends that police departments retain data for seven days, reduced from the former 30-day period. In conjunction with this revision, Flock has rolled out “Evidence Mode,” intended for cases that necessitate extended data retention and require a specific case number. Furthermore, customers have gained the ability to manage the data shared with others depending on the offense being investigated.

Additionally, Flock has unveiled a feature named “Audit Assistance,” which was launched in April. The company indicates that more than a third of its users have activated this tool, which will become mandatory for all customers by the year’s end.

Flock states that Audit Assistance “detects unusual activity and flags it for Administrator review,” as highlighted in a blog post released on Thursday. When flagged, the tool can automatically suspend user access until an administrator intervenes. Flock asserts that this feature has already helped customers identify and prevent misuse.

Though numerous executives have lauded Audit Assistance, Flock has not yet provided a comprehensive explanation of how it operates.

Some comments shed light on Audit Assistance’s capabilities, but specifics about its functionality remain unclear.

For instance, Ashley Haber, Flock’s head of trust and compliance, remarked in a video that the tool can flag “potentially unusual search histories from particular users,” allowing customers to assess how their users engage with the system. In another video, Flock co-founder Paige Todd explained that “Audit Assistance swiftly flags irregular patterns, such as a user repeatedly searching the same license plate for more than 30 days.”

Major Patrick Krieg from the Dunwoody, Georgia Police Department, a Flock customer, described the feature as “an algorithm that alerts us to any signs of bias.”

Flock depicts Audit Assistance as a tool capable of “early identification of atypical activity, allowing prompt assessment through documented workflows.”

Nonetheless, it is still uncertain — and TechCrunch has sought clarification on — how this “algorithm” works, the data it uses for training, and whether it was built using a large language model. TechCrunch has also asked about the specific patterns it aims to identify and what defines “abnormal activity.”

Flock has clarified that Audit Assistance does not use [Machine Learning] or AI. The company informed TechCrunch that “this is a data tool that flags unusual search behavior for further scrutiny. For example, if a user repeatedly searches for the same license plate using various case codes, it could signal misuse. A flag simply indicates that the behavior deserves a closer examination.”

Still, many questions persist, especially concerning any statistics or data Flock has that demonstrate the tool’s effectiveness, its rates of false positives and negatives, and whether it has undergone third-party auditing.

Privacy advocates and critics of Flock have expressed skepticism about Audit Assistance.

Chad Marlow, senior policy counsel with the American Civil Liberties Union, commented on Flock’s announcement, stating that “there is no proof that the tool reliably addresses misuse.”

“Despite numerous officers recently facing arrest, dismissal, or other disciplinary actions for improper use of Flock data for personal purposes, Flock claims that these incidents verify the effectiveness of its auditing tool,” Marlow noted. “However, without knowing how many officers have misused the system, we cannot determine if Flock’s audit tools catch 95% or merely 5% of violators. The auditing tool should be evaluated by an independent assessor to validate its true efficacy. Until then, it’s unclear if it is a valid security measure or just for appearances.”

Cooper Quintin, a security researcher and senior public interest technologist at the Electronic Frontier Foundation, told TechCrunch that the effectiveness of Audit Assistance may not be the main issue.

“If it is effective, law enforcement will likely find ways to bypass it,” he suggested, noting that accountability is the ultimate concern. “If there are no consequences for misuse, and the entity analyzing the reports is the same police force committing the offenses, it merely serves as a façade for Flock. The best way to ensure accountability is to implement laws regulating the use of this technology and requiring a warrant for its application.”

Rachel Levinson-Waldman, director of the Brennan Center’s Liberty and National Security Program, shared with TechCrunch that mandating Audit Assistance is a positive step and that “in theory” the tool provides a “great concept.” However, she stressed that the tool needs to be audited to clarify how it functions in practice.

Flock’s announcements come in light of several reported incidents involving customer misuse of its technology. Earlier this week, three former deputies from a Georgia sheriff’s office were arrested for allegedly using Flock cameras to stalk individuals with whom they had prior personal relationships, as outlined in a release from the Bibb County Sheriff’s Office. In this context, Sheriff David Davis specifically acknowledged Flock’s new Audit Assistance tool for exposing the misconduct.