Flock AI example shows municipalities need governance help

Flock safety has been in the news a lot lately and is a great example of cities not understanding what AI can do inside tools and not having the right questions to ask before spending taxpayer dollars and political capital on something that may bite them in the ass.

Flock Flock Safety sells automated license plate reader cameras, often called LPRs, to police departments, cities, neighborhoods, HOAs, businesses, and property owners. These cameras capture license plates and vehicle details, then compare them against lists such as stolen vehicles, wanted vehicles, missing persons alerts, or vehicles connected to investigations. Flock says its LPR data is retained by default for 30 days, then hard deleted from the cloud.

Over 5000 municipalities across the country have adopted flock. On the surface the argument is compelling. If you need to find a missing persons or an active shooter the technology is great. The concern however is the ever growing presence of a surveillance state that these cameras are helping to usher in.

Flock and other AI tools are powerful public safety tools that should only be used under a public governance framework.Of course there is a blessing and a curse to most things however many of these municipalities are bringing on Flock for the same reason Jnco jeans were popular in highschool for me, because everybody else is doing it! Decisions are being made without intense knowledge of the pitfalls and risks to populations. The rest of the time here we will talk about what municipalities should be thinking when bringing on new technologies using Flock as the example. A powerful public safety tool that should only be used under a public AI and surveillance governance framework

Let me start this by saying Flock is like every other tool. It is neither good nor bad. The way flock or the way the city uses the information gathered is what is good or bad and that can only be judged by the observers. This podcast is to increase the observers. For you in municipality governance, here is a framework you should copy from Janis Consulting to make decisions when it comes to technology you don’t understand.

1) Procurement and approval

-AI tools are inherently different than other tools brought on by city departments as they can have cascading privacy effects. A police department should not be able to bring on a tool like flock without securing approval from the cities AI chief officer. If the city doesn’t have one a consultant that understands all sides needs to be consulted.

Flock adoption, unlike Jnco jeans, has huge privacy, political, and safety ramifications. These all need to be understood from all angles before making decisions that impact the cities population. Lastly, narrative control to make the adoption a good thing instead of a shit show to be used later politically. Organizations do a poor job of this as a whole.

2) Understand Data retention rules

-Flock has a general 30 day data retention policy. This however can be negotiated which many people do not know. Additionally, understanding the process helps greatly. Should routine, non hit plate leads be deleted sooner as an example? This prevents a tool meant for “finding stolen cars” from becoming a long-term movement database of ordinary residents.

3) Data Sharing controls

Flock’s value grows because agencies can potentially share data across jurisdictions. But that also creates the risk of a local city accidentally participating in a national surveillance network. Which agencies can access the new information? Where have other municipalities had success sharing and had problems sharing? Can private cameras feed into public police searches? Have somebody with knowledge of what is possible be in your corner.

4) Understand search an use restrictions

-The city should define allowed uses and banned uses based on its population. Chicago is going to be different than Dallas than IDK Omaha. Perhaps you want to use it to find stolen cars but not use it to see who went to a certain political rally.

This is not theoretical. San Francisco recently cut off access after nearly 300 improper Flock searches were conducted on behalf of federal and out-of-state agencies, which violated California law.

5) Human in the loop

-This to me is one of the biggest and takes the most time to get right. With flock, is a LPR enough to pull over a vehicle? What happens in the case of mis-reads? Should there be human verification before action is taken and where in the process?

Those that watch this channel know that I believe AI will be a great tool to propel humanity as long as we do it in a human first way. Humans have wisdom and it needs to be baked into every equation using these tools.

Work with the people on the ground and learn from them when the tool can be most effective.

6) Audits, transparency, and accountability

-What gets reported? Who is in charge of reporting? How often does the “dashboard” of information get updated? Who inside the organization keeps feeding the accountability fire?

In the case of flock, all these topics should be managed:

Number of plate reads

Number of hits

Number of searches

Number of investigations supported

Number of arrests or recoveries

Number of false positives

Number of outside agency requests

Number of policy violations

Data-sharing partners

Any complaints or lawsuits

Camera locations added or removed

7) Last but not least, cost.

Flock operates on a per/ camera subscription model. Depending on the contract 3,000 / camera/ year. Additionally, ~$700 for each camera installation. Depending on the size of the city this adds up to a pretty good chunk of change. This is only 1 AI tool. Many companies operate on a similar idea.

Cities should have an idea of what is most important to their constituents and consult professionals to understand what tools work best to solve those problems first.

Governance is the most important part of AI. Let’s work together to make sure your organizations governance is top notch.