Ulster bans its county from using Flock cameras and AI surveillance: a limited ban that sets the local course

By Daily Freeman · October 22, 2026.

Ulster County (New York) Executive Jen Metzger signed an executive order on Thursday banning the county from using any Flock Safety services. It also bars “similar AI-assisted technologies,” explicitly including vehicle or person re-identification. According to her office, the order takes effect immediately.

Its scope is narrower than the headline suggests. According to the Daily Freeman, there are currently no Flock cameras in the county. The City of Kingston uses automated license plate readers (ALPR) to enforce violations in four school zones, but those are maintained by the German company Jenoptik and reviewed by the city's Parking Violations Bureau. Nor does the order apply to the sheriff's office or to any agency outside the executive's authority. That office has used Motorola readers for more than a decade and, according to the paper, has said it does not and will not use Flock.

Sheriff Juan Figueroa backed the order in a statement issued by Metzger's office. He said it provides “an important framework” for protecting public safety without sacrificing privacy and civil liberties. Related articles linked by the paper indicate that the sheriff had already dropped a plan for AI-assisted reporting and the installation of more license plate readers. Those pieces are not detailed in the material I was able to review, so I cannot specify the sequence or the reasons.

Metzger justified the measure on political grounds: AI, she said, endangers “the rights on which this county was founded 250 years ago.” She spoke of a shift “from passive technology to active surveillance.” She added that the private AI sector has no “regulatory guardrails” and that she is concerned about “very troubling” security breaches. These are her statements, not facts verified by the paper.

**Our take.** The order is largely preemptive. It bans something the county does not use, which is why it carries little political cost. But that is precisely its value: it draws a line before any technological dependence exists, while giving it up is still cheap. Banning a system already embedded in a police department, with a budget and established habits behind it, is much harder.

What matters is the approach. For context on the sector, in the absence of clear federal rules on AI surveillance, regulation is being improvised by counties, cities and police departments. That pattern tends to produce a patchwork, with rights that vary by ZIP code. There is an upside, because it serves as a policy laboratory. There is also a downside: an executive's ban can be revoked as easily as it was signed, and it leaves out anyone not under that executive's authority, as with the sheriff here.

It is also worth separating two things the debate often conflates. Reading license plates to issue fines in a school zone and re-identifying people across cameras are not the same. The second turns a passive camera into a system that tracks individuals, and that is the qualitative leap Metzger describes. A good rule should distinguish purpose, data retention periods and judicial oversight instead of banning “AI” wholesale.

This fits our usual thesis: regulation should be grounded in evidence and in concrete capabilities, not panic. In the short term, automated surveillance is one of AI's most tangible risks: power can concentrate without any superintelligence. In the long term, a society of abundance is only desirable if civil liberties remain intact. Setting those limits now, while the technology is still optional, is a cheap investment in that society.

What to watch is whether the order becomes more durable rules, such as an ordinance passed by the county legislature, and whether the sheriff turns his support into concrete public policies. Metzger said she is working with him to define them but has not detailed timelines or content.