Police AI Tool Expands Surveillance Beyond License Plates

Police AI Tool Expands Surveillance Beyond License Plates

2026-08-20 companies

Atlanta, Wednesday, 19 August 2026.
Surveillance firm Flock Safety is facing intense scrutiny following revelations that its next-generation artificial intelligence system, OS Investigate, extends far beyond simple license plate reads. Codebase analysis reveals the platform cross-references real-time camera data with commercial databases—including Social Security numbers and home addresses—to track driver movement patterns and identify personal associates. Processing over 20 billion monthly plate scans across 6,000 communities, the $7.5 billion company allows police to run automated, warrantless profile checks. Amid mounting privacy lawsuits and widespread reports of officer misuse, Flock Safety introduced security guardrails in August 2026, including shortening default data retention from 30 days to seven. However, civil liberties advocates argue these self-imposed measures fail to address core constitutional concerns regarding mass automated tracking.

Advanced AI Capabilities Revealed

The core of the controversy lies in the technical specifications of the new system, known as OS Investigate. A reconstruction of the platform’s codebase indicates it utilizes 69 prewritten prompts and accesses 45 distinct AI tools to process data [1]. Unlike previous iterations limited to license plate recognition, this system integrates real-time camera feeds with commercial databases containing Social Security numbers, dates of birth, and home addresses [1]. The software allows law enforcement to track vehicle movement patterns and identify potential associates based on proximity and frequency of movement over periods such as the 14 days prior to a query [1]. Flock Safety states the product is currently in development and undergoing testing with a small group of law enforcement partners, though capabilities may change before commercial release [1].

Policy Adjustments and Data Retention

In response to growing scrutiny, Flock Safety announced significant updates to its privacy and data retention policies on Thursday, August 13, 2026 [2]. The company shortened the recommended default data retention window from 30 days to seven days, representing a 76.667 reduction in the standard storage period [2]. Additionally, an “evidence mode” was introduced to allow detectives to preserve specific automated license plate reader data in exceptional cases [2]. To address reports of law enforcement misuse, the company is mandating an audit assistance feature and a proactive lockout feature that suspends access if user activity exceeds an abnormality threshold [2]. These guardrails were confirmed via social media announcements shortly after the policy update [4].

Legislative and Public Backlash

Political pressure has intensified alongside the technical revelations, with bipartisan concern emerging at state and federal levels. On August 6, 2026, two Connecticut Senate leaders officially called for a pause on the installation of new Flock cameras until the end of the 2027 legislative session [3]. Governor Ned Lamont subsequently requested the Connecticut Police Officers Standards and Training Council review automated license plate reader technology and develop privacy guidance within 30 days of August 9, 2026 [3]. This legislative action follows incidents including a Texas officer conducting over 83,000 searches in 2025 to track a woman regarding a self-administered abortion [1]. At least two bills have been introduced in the U.S. Congress during 2026 to restrict the technology’s usage, though their status remains pending [2].

Future Outlook and Accountability

The trajectory for surveillance technology regulation remains uncertain as agencies balance public safety with privacy rights. Flock Safety announced on August 11, 2026, that by the end of the year, it will implement monitoring for abnormal searches and mandatory case code attachments [1]. Meanwhile, Connecticut’s artificial intelligence caucus seeks to establish clear rules for organization and monitoring by late 2027 [3]. Privacy advocates argue that self-regulation measures may be insufficient without statutory limitations on warrantless searches [2]. The company continues to hire engineers for “agentic policing” systems, with completion expected by the end of 2026 [1].

Sources


Artificial Intelligence Surveillance Technology