What Changed on August 13
Flock Safety's own trust and compliance pages, updated August 13, 2026, and CEO Garrett Langley's public statement the same day set out four changes. Default license-plate data retention falls from 30 days to 7 for agencies that do not choose otherwise. A new Evidence Mode lets agencies keep specific footage longer when an active case needs it, offered free to agencies that adopt the shorter 7-day default. A misuse-detection audit tool that flags abnormal search behavior, until now optional, becomes a mandatory requirement for every law-enforcement customer by January 1, 2027. And a new proactive-lockout feature will automatically suspend an officer's account when a search pattern matches known abuse signatures, pending human review.
The Verge's Lauren Feiner, CNN Business and an AP wire report all covered Langley's statement on August 13, in which he described the recent pattern of misuse as 'completely unacceptable' and said the company was assuming greater responsibility for how agencies use its network, rather than treating misuse purely as a customer-side policy failure.
The Misuse Cases Behind the Reversal
The reforms follow a run of publicly reported abuse cases in which police officers used Flock's license-plate network for purposes that had nothing to do with a criminal investigation, including searching for the location of romantic partners or former partners. Several of the officers involved have since resigned or been arrested over the conduct, according to the same August 13 coverage.
Those cases are the ordinary failure mode critics of police surveillance technology have warned about since the network's earliest rollouts: a tool built to search vehicle locations at scale is only as accountable as the audit trail behind each individual search, and until this reform that audit trail was optional rather than mandatory for the agencies actually running the queries.
Self-Regulation as a Pre-emptive Strategy
Langley did not stop at fixing Flock's own defaults. He also called on state regulators to mandate misuse audits and impose stiffer penalties across the industry - an AI vendor asking to be regulated rather than waiting to be. That is a notable sequence: converting an emergency policy fix into a public floor before a regulator can impose a stricter one, and doing it in a way that puts pressure on any competitor still running an optional audit tool.
The European Union's own framework already treats this category of technology as consequential: the EU AI Act classifies AI systems used by law enforcement for tasks like this as high-risk, with logging and human-oversight duties attached. A US vendor voluntarily moving from optional to mandatory misuse auditing is, in effect, a preview of the operational floor that high-risk classification is designed to eventually require of comparable systems operating in the EU and UK.
What This Means for Anyone Running Monitoring AI
The specific number to remember is not the retention window, it is the date: January 1, 2027 is when an audit tool that used to be optional becomes the floor Flock's own customers must meet. Any organization operating or procuring a system where an individual's search activity can be turned against a private person - not just license-plate networks, but workplace-monitoring tools, retail surveillance systems, or internal search logs over customer data - now has a public, named benchmark that a regulator, an auditor, or opposing counsel can point to.
Treating an abnormal-behavior audit log and an automatic lockout as optional add-ons was a defensible design choice before August 13, 2026. After a named vendor has publicly reversed that position following real misuse cases, it is a harder position to defend, in the EU and UK as much as in the US, given how squarely comparable systems already sit inside the AI Act's high-risk category.
Read next: Meta Paired an AI Manifesto With a Data Center Pitch | It Was Told to Book a Gym Class. It Hacked the Booking System Instead.
