Thu 13 Aug 2026 / 14:49 ET
Kernel
AI 3 min read

Flock Safety announces new search rules and shorter data retention

Flock says it will require case codes and automated audit tools as critics question whether the new controls curb police misuse.

Theo Lindgren

By Theo Lindgren / Columnist

Flock Safety new search rules will make officers tie searches to a case code and put an automated audit tool in place, the company announced Aug. 13, as scrutiny grows over alleged personal misuse of its license-plate-reader network. The controls were previously optional. Flock CEO Garrett Langley told the Associated Press they are due to become mandatory by Jan. 1, so this is an announced rollout, not proof that every agency is already operating under the new rules.

Flock’s cameras record license-plate data and other vehicle characteristics, according to the AP, and agencies that contract with the company can search and share that information across jurisdictions. Police and Flock describe the system as an investigative tool. Civil-liberties groups see a nationwide system capable of exposing people’s movements at uncomfortable scale.

What will Flock’s new search rules do?

Under the announced policy, a law-enforcement user will have to enter a code from a records-management system that links a search to a specific case. An emergency override will remain available, but Flock says those searches will be flagged for review.

Flock also says it will require agencies to use its Audit Assistance feature. The tool analyzes search activity for abnormal patterns and can lock a user out until an administrator reviews the account. That is a useful escalation over a box nobody has to tick, though it remains an internal administrative process rather than outside review.

MIT Technology Review reported that Flock will not verify the case numbers entered by officers. The same report said the company has not disclosed the audit tool’s accuracy or allowed independent evaluators to test it. Those limits matter because the case-code control depends on an officer entering a real connection to an investigation.

How long will Flock keep license-plate data?

Flock says it is changing its standard retention period from 30 days to seven days. Data tied to a case number may be kept longer as evidence, and agencies may have longer periods where local law requires it. MIT Technology Review also reported that agencies can override the seven-day recommendation, meaning the shorter period is a new default, not a universal deletion deadline.

Customers will also be able to limit what other agencies may search in their data by offense type. The AP reported that a department could use that setting to block outside searches for immigration enforcement while allowing searches for specified crimes. The choice still belongs to the customer agency.

Why is Flock changing its policies?

The announcement follows reports of officers accused of using Flock searches for personal purposes, including stalking. MIT Technology Review cited a Washington Post investigation that identified 46 cases involving accusations of unauthorized use. The AP separately reported that critics and lawmakers have raised concerns about access to the data, including for immigration enforcement.

There is a contract backlash, too. The AP cited a DeFlock tracker showing more than 50 agencies or jurisdictions had canceled, suspended or rejected a contract, or deactivated cameras, since the start of the year. That tally covers several different actions, not just cancellations.

The ACLU called the shorter retention period a possible step forward but described the other changes as recycled, inadequate protections, according to the AP. Institute for Justice attorney Robert Frommer said searches should be governed by judges and warrants rather than police discretion. Langley has said Flock is shifting from offering optional safeguards to enforcing guardrails. The company has changed the defaults; the dispute is whether that changes who has meaningful control over a search.

This story draws on original reporting from MIT Technology Review.

More AI/

view all ↗