Someone dressed as Darth Vader stood up at a San Diego public meeting and testified in favor of the city's license-plate cameras. "The emperor is a fan of Flock," the speaker said, "and we must continue utilizing Flock technology" — so the empire could follow the rebels from playground to playground, from pool to gymnasium. The cameras, he went on, were not only reading plates; they were following children, in parks and gymnasiums. He finished by saying he needed the system so he could stalk an ex-girlfriend.
The clip opened a segment on Moonshots with Peter Diamandis about Flock Safety, a public-safety technology company founded in 2017. Its core product is a network of AI-powered cameras that read license plates for police departments, neighborhoods, schools and businesses, to identify vehicles connected to crimes and missing people. The company has added video cameras, gunshot and audio detection, mobile surveillance trailers, drones and software that ties them together. Flock says its systems are used in more than 6,000 communities across 49 states and generate over 20 billion vehicle scans a month. Flock's own explanation of its technology reports the same figure; it is a count of observations, not of distinct vehicles, so one car passing many cameras is counted many times.
The first AI a town meets
Alex, who brought the story to the show, said this is what AI looks like at the municipal level. Discussion of AI usually runs to enterprise software and scientific discovery, he said, but for many communities the first encounter will not be a driverless car. It will be a camera on a pole reading plates. He said he had read reports of people chopping down cameras from Flock and similar companies because they felt surveilled by their own police department, and reports of police and other first responders using their access to track ex-partners — the behavior the costumed speaker was mocking.
His larger worry was federation: police departments sharing their camera data with one another, so that a search in one town can follow a car across many. That, he said, turns the question from what the cameras record into who is permitted to look — who gets to see whom inside the panopticon, as he put it, crediting the framing to David Brin. In China, Alex said, it is obvious that the party looks through the cameras at everyone else. In the United States the answer is still being argued, which is why he called Flock a petri dish for the question. An empire uniform testifying for surveillance in front of a city council struck him as a very American way to have it.
Flock describes controls on the looking. The company says customers own their data and must explicitly authorize sharing, and that private customers cannot reach government data. Searches, it says, require a documented case-related reason; audit trails are kept indefinitely; plate observations are deleted after 30 days by default, or according to applicable retention laws. It points to public transparency portals as a way for residents to inspect how their local agency uses the system. Those are the platform's stated rules.
From finding a suspect to finding people who look like one
Salim Ismail said the structural change is not the camera but the economics. "We've transitioned from find this suspect to find me people who behave like suspects," he said, "and that's a very, very different animal."
His argument is about friction. Following everyone used to be limited by manpower: an investigation required officers, hours and a reason to spend them. Pattern matching across a federated camera network costs almost nothing per query, so the constraint disappears. "Civil liberties were kind of protected to some level by the friction, because following everybody was just too expensive," he said. "And now that's gone." His summary: "The Constitution does not update when transaction costs go to zero."
He pointed to misuse already reported — officers using the tools on partners and ex-partners — and said people had died after being tagged with the wrong plate because a reader misread it. Misreads are a known failure of the technology; Flock's explainer is written around what investigators do when plate data fails, describing how its systems try to identify a vehicle by other attributes when a plate is missing, altered or read incorrectly.
Ismail said the American arrangement is converging on the Chinese one from a different direction: there the party mandates it, here it is assembled by corporations and by police forces he described as operating with little warrant or oversight. He wanted the argument raised to the constitutional level, and said he did not think it ends well.
The other side of the trade
Diamandis offered the strongest case for the cameras. It is not only China that runs this way, he said; the Emirates do too. He described a conversation with Jamie Justice, who heads life sciences at XPRIZE and is a prolific runner. She had gone out for a run at ten at night in the streets of Dubai, something she said she would never do in the United States, and felt completely safe.
There is always a trade, Diamandis said, and he makes it routinely: facial detection at airport customs moves him through instantly instead of leaving him in a long line, and he will take that deal to save time. He would take it again for his children's safety. His open question was procedural — whether such a trade belongs to each individual or to a society deciding collectively that it values safety over privacy.
Alex answered with Benjamin Franklin: those who would give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety. Ismail agreed, and said the quote should sit on every desk for the next decade.
Ismail also qualified the Dubai example rather than denying it. The UAE, he said, is about as benevolent a dictatorship as one could imagine, and that benevolence is exactly why people feel so safe: step out of line and you are removed quickly. The structural problem is that benevolent dictatorships typically do not stay benevolent for long. The systems outlast the intentions of the people who installed them.
The Afghan warlord, and how long a mistake takes to fix
Ismail's own story is about what happens when an automated match points at the wrong person. Arriving in the United States, he was pulled aside and told he would miss his flight, because an Afghan warlord wanted by the FBI for poppy trading had a name close to his own. The officers gave him the good news first: they were fairly clear that a vice president of Yahoo who speaks at conferences was not the warlord. The bad news was that they were not allowed to make that call themselves. Washington had to confirm it. Have a seat; it will take a few hours.
It took a few years to clear up. When he asked whether every Salim Ismail was stopped this way, he says an officer replied that when one Juan Rodriguez commits a misdemeanor, every Juan Rodriguez gets pulled aside. Ismail was careful about the people enforcing it: the U.S. immigration officers he has met have been, in his words, nothing short of "beautifully professional," to the point where after enough crossings some knew him by name and still had to run the check.
That is the analogy he brought to the cameras, not a Flock case. Technology can fix this kind of problem, he said — names can be disambiguated quickly and resolved precisely. The same technology, applied without that care, produces automated suspicion that is very hard to argue your way out of, and he described the presumption of innocence and habeas corpus as badly eroded as a result. Getting caught on the wrong side of an automated match, he said, can go very badly for quite a long time.
Pointing the cameras back
Alex raised the remedy he associated with Brin and others: rather than restricting the cameras, give everyone access to them. Let civilians run the same plate searches that police and first responders run. If it happens in a public space, make it a public resource — let everyone track everyone.
The answer from the panel came with praise for the Electronic Frontier Foundation's work on these questions, and with Russian dashcams as evidence that citizens do point cameras back when they have reason to. Russian drivers adopted dashboard cameras early, the argument went, as protection against police who would break a headlight and then extract a bribe over the ticket. The side effect showed up when a meteor came down: suddenly there were something like 200 live videos of it, not because that many people had whipped out their phones in time, but because the cameras were already running. The citizenry, on this reading, is starting to point the camera the other way — "and we need a ton more of that."
The proposal does not reduce the watching; it distributes it. Opening plate searches to everyone would mean any resident could look up where a neighbor's car has been in public, alongside any officer who can already do so. That is the trade at the center of Brin's idea, and on this episode it remained a proposal rather than anything a city has adopted — offered against Ismail's darker reading, in which the cost of following everybody has already fallen to zero and the rules have not caught up.