Flock Safety’s automatic license plate readers were first sold as upmarket add-ons for gated communities and HOA curtain-twitchers who wanted to keep tabs on what everyone was doing, whether they lived there or not.
Flock soon realized that selling to rich people wasn’t as profitable as selling to governments. Law enforcement agencies have always been surveillance tech early adopters — something that springs equally from being able to spend other people’s money and a desire to get out ahead of constitutional case law.
So, now Flock is pretty much everywhere. But instead of getting better due to constant refinement, it seems to be content to rest on its ubiquity. If you’re already everywhere, why try harder? After all, ending a contract with Flock doesn’t mean the company won’t keep your cameras activated so other law enforcement agencies can access plate/location information. And if federal agencies are using local agencies to bypass restrictions on plate reader access, well… that’s just an end user problem.
But if you buy license plate readers, you kind of expect them to read license plates. Well, that’s not happening in Roseville, California, where Flock’s cameras are about as effective as pole-mounted Polaroids:
Flock says that in optimal conditions, its cameras accurately read more than 96% of license plate characters. Hundreds of pages of records from the Roseville Police Department show a different picture. In 2023 and 2024, Flock sent 1,427 alerts to Roseville police, flagging vehicles as stolen or used in a felony after they passed one of the city’s Flock cameras. An analysis by the police department found that in 71% of those alerts, Flock’s machine-learning software incorrectly read the license plates.
That is objectively terrible. And because it is so terrible, Flock’s flack has decided to blame the end users.
A factor that contributed to the misread problems in Roseville was the “particularly unique deployment” that the city requested, a Flock spokeswoman said. Roseville said it has its cameras configured so that they capture only the backs of vehicles, a setup intended to avoid capturing personally identifiable information like faces. Roseville’s setup included older hardware and placement of cameras higher and further from vehicles than the company typically recommends, Flock added.
Huh. Well, that might mean something if Flock had made any honest effort to make this “unique deployment” work better. But it doesn’t appear to have done that. Instead, it simply insists things are better now — a statement that’s not backed by any evidence Flock’s has on hand and/or is willing to share.
Flock said Roseville’s camera performance has significantly improved, which the police department disputed.
According to emails obtained by Business Insider, it does appear that someone appears to be working on the ongoing issue, albeit at a leisurely pace. This is the email sent to Amanda Jones, Roseville PD’s real-time crime center supervisor by an unnamed “Flock Senior Product Manager.”
I 100% understand, and it pains me to no end that we are missing reads for you. I am grinding on this issue, and am writing code as we speak. I am improving systems at Flock internally that can automatically identify when we miss cars and then diagnose the issue so that we do not need to rely on customers to tell us.
I think it may be helpful for you and I to have a conversation so that you can get a view under the covers of what we are working on to improve this?
We have an all company summit during the 2nd half of this week, but if you would like to discuss later this month or next, please let me know and I can share my availability.
While I understand that Flock’s customer base means product managers are stretched pretty thin, we’re not talking about malfunctioning doorbell cameras or warranty repairs. We’re talking about tech that is capable of depriving people of their freedom, if not their actual lives, when it’s wrong. That’s the sort of thing that shouldn’t be met with vague promises to talk about this long-known, ongoing problem at some point in the next several weeks. That Flock doesn’t treat this as a priority shows it doesn’t care about anything more than selling cameras and raking in service/access fees.
Fortunately for residents of Roseville, Flock being wrong hasn’t led to police stops or wrongful arrests. That’s because the PD forces officers to personally verify plates and cannot use a plate read alone to justify a traffic stop.
But most places using Flock cameras don’t do that. Fewer still place any limits on access to the company’s database of billions of plate/location photos, which means cops from all over the nation are using junk data generated by faulty cameras to initiate stops and pursue investigations.
And it’s not just the false positives. Part of the PD’s frustration was Flock’s inability to generate records when officers needed them most. In at least two cases, vehicles driven by criminal suspects managed to pass through the network of Flock cameras without generating a single usable plate photo. Meanwhile, the cameras were turning 2’s into 3’s, N’s into V’s, and apparently just guessing when the cameras failed to capture an entire plate, rather than simply discarding the incomplete read.
Again, this doesn’t appear to matter to Flock. It will oversell its capabilities and blame everyone but itself when things go wrong. It has already shown it’s unwilling to improve until federal oversight starts demanding answers. And this approach to testing shows it’s not serious about actually improving its products:
In 2021, the research firm IPVM independently tested Flock’s license plate readers, concluding that Flock misidentified the state in about one of 10 reads, and that the system regularly misclassified vehicles’ type and make. IPVM said that Flock subsequently blocked it from purchasing its cameras for testing.
What’s weird about this is that the city of Roseville says it’s going to keep throwing money at Flock, despite its extensive record of failure. This makes it feel like less of a tech company and more like a cult. The feeling seems to be that if the cameras ever generate even a single arrest, they’re worth paying for, even when they’re wrong most of the time. And that’s something that can’t be fixed simply by changing surveillance providers.