WIRED Reconstructs Flock Safety’s ‘OS Investigate’ AI Surveillance Tool
The reconstructed software allows officers to search by physical description and behavioural patterns, raising constitutional concerns among privacy experts.
A reconstruction of the code for Flock Safety’s ‘OS Investigate’ AI system has revealed capabilities that extend far beyond traditional license plate recognition, enabling police to track individuals and identify associates based on movement patterns. The software, currently in development and tested with a limited group of law enforcement partners, allows officers to search for people using physical descriptions, location data, and behavioural indicators without requiring a specific name or vehicle plate number.
The system integrates camera metadata with commercial identity records, arrest files, 911 dispatch logs, and ballistics results to generate detailed dossiers on individuals. WIRED accessed the software interface by loading Flock’s login pages, which served a cache of more than 450 files containing code for 45 tools. This allowed reporters to reconstruct the authenticated application without actual police access, uncovering 69 prewritten prompts that officers can select or edit to initiate searches based on behavioural patterns rather than specific incidents.
Privacy experts and legal analysts have raised concerns about the constitutional implications and potential for warrantless surveillance. The code reveals that the system can strip out buses, semi-trucks, work vans, and trailers from search results, potentially flagging ordinary delivery drivers while excluding commercial transport. A default ‘confidence threshold’ of 0.75 is used to identify ‘associates’, defined as vehicles appearing at the same cameras within a two-minute window of a target vehicle three or more times.
Flock Safety maintains the tool is designed to assist investigators using data agencies already possess, describing it as a separate product from its license-plate reader technology. The company announced plans to implement new safeguards by the end of the year, including monitoring for abnormal searches, automatic lockouts for flagged users, and mandatory case codes. However, the search justification form defaults to requiring a typed reason but places no strict requirements on the content or length of the justification.
Flock is hiring engineers to expand the system’s architecture to include automated lead generation and cross-camera correlation, moving beyond the version examined by WIRED. The company faces bipartisan political pressure and a growing record of officers misusing its platform, with privacy advocates arguing the tool casts suspicion on drivers based on where and how often they drive, regardless of whether they are wanted for a crime.
