Brian Reed: Hey. You see both of these drop?
Eliza Ward: NPR yesterday, WaPo the day before — yeah.
Brian Reed: So NPR's number is roughly 100 cities — deactivated, rejected deals, canceled contracts, all tracked through DeFlock's database. That's the confirmed figure as of August 22nd.
Eliza Ward: And this is a company a16z and Tiger Global backed at eight billion dollars, deployed in over five thousand communities across forty-nine states. So — wait, a hundred cities is what percentage of that?
Brian Reed: That's the part I don't get. Is a hundred cities a crisis or just noise at that scale?
Eliza Ward: And the WaPo piece — August 21st — says this is going bipartisan. That's the detail that actually made me stop.
Brian Reed: Right — because if it's just progressive cities pushing back, a16z can probably wait that out. Bipartisan changes the math.
Eliza Ward: If it holds. We don't know yet if bipartisan framing translates to bipartisan votes on anything.
Brian Reed: That breaks the 'hundred cities' frame though — there are still four thousand nine hundred communities running this. And I keep thinking about what the system is actually doing at that scale, because calling it a camera company is like calling Google a search box.
Eliza Ward: Okay, put it plainly.
Brian Reed: Every road you drive, every stoplight, every intersection — imagine it photographs your license plate and sends it to a private company's database. Not once. Every time. That's Flock. Twenty billion captures a month.
Eliza Ward: Twenty billion — per month. And the Institute for Justice lawsuit actually has language for this. They argue it is, quote, functionally impossible to drive without being recorded and stored in Flock's AI-assisted database. That's the lawsuit's framing, not ours.
Brian Reed: So the hundred cities canceling — I mean, the math doesn't close, does it? Like, if it's actually impossible to avoid being scanned, then whether your city opted out... might not matter much to your actual data.
Eliza Ward: Right, and that's exactly what the ACLU and EFF are flagging — it's not just one city's choice anymore, it's a network effect. Matthew Guariglia at EFF has been pretty direct that his objection is to ALPR surveillance broadly, not just Flock's version of it. The infrastructure itself is the problem.
Brian Reed: So the retention cut — 30 days down to 7 — does that actually touch what Guariglia's describing? Because if the scan volume doesn't change, seven days of a twenty-billion-scan-a-month ledger is still... a lot of driving history.
Eliza Ward: That's the unconfirmed part. Who actually audits whether Flock follows the new retention rules? Customers? No regulator's named. So we have a policy announcement — and no enforcement mechanism we can point to yet.
Brian Reed: And the audit gap is exactly where the 'abstract privacy concern' framing falls apart. Because — look, people keep saying this is a philosophical debate about surveillance. But NPR's August 22nd report names police officers using Flock data to stalk women. That's not a hypothetical. That's documented. That happened.
Eliza Ward: Specific officers. Specific women. Named in reporting.
Brian Reed: And then Texas — a deputy used Flock data to track a woman's movements, matched her route to a known abortion clinic, built a case that way. So the same camera network catching car thieves is now — I mean, it's doing reproductive policing. Not in theory.
Eliza Ward: Wait, and there's also the accuracy problem sitting underneath all of this. Chrisanna Elser — Colorado woman — accused of package theft because Flock and Ring cameras put her car near the scene. Police called it, quote, incontrovertible. And it was disputed. So you've got the abuse cases, and then you've got cases where the system just... gets it wrong and ruins someone's Tuesday.
Brian Reed: Which connects directly to the Amazon Ring partnership — Flock clients can request Ring footage from private homeowners. So now disputed evidence chains through two private companies before anyone asks a judge.
Eliza Ward: No, I don't buy that the guardrails touch any of this.
Brian Reed: The stalking? The abortion investigation? Seven-day retention doesn't un-happen those. The data was used within days. Probably within hours.
Eliza Ward: And the immigration sharing — Flock data went to federal agencies without users' knowledge, per NPR. That's not a retention-window problem. That's a policy problem nobody's confirmed is actually closed.
Brian Reed: Right — and the part that matters later is whether a16z and Tiger Global can hold an eight-billion-dollar valuation once those documented harms start attaching to actual legislation. That's the question we haven't gotten to yet.
Eliza Ward: That legislation question — let me actually pin it down, because 'valuation at risk' is one thing, but what breaks the model specifically? Flock cut from 30 to 7 days only after Denver and Longmont deactivated cameras. That's the sequence. The guardrails didn't come from principle. They came from Denver pulling the plug.
Brian Reed: And agencies can still modify the retention period. So the seven days isn't actually a floor.
Eliza Ward: Right — it's a default. A starting point that any police department can walk back in coordination with state or local policy. So the announcement is... I mean, it's minimum-viable-concession language dressed as reform.
Brian Reed: And who confirms they're actually following even the default? Like — Flock's customers? There's no named regulator.
Eliza Ward: This is me speculating now — but if Flock's real asset is training AI models on scan data, seven days might be sufficient. You don't need thirty days of a parking-lot attendant's Tuesday to train a vehicle-behavior model. You need volume. And they have twenty billion scans a month.
Brian Reed: Hang on — so the retention cut might not even cost them the thing they actually care about.
Eliza Ward: Unconfirmed. But it's the question nobody's answered. And then — Garrett Langley, Flock's CEO, has said the platform is simple and customer-controlled. The contract language apparently tells a different story, though I don't want to overstate what's confirmed there.
Brian Reed: The investor math matters. A16z, Tiger Global, Kleiner Perkins, Founders Fund — over a billion dollars in, valuation somewhere between eight and eight-point-four billion. A hundred cancellations against five thousand communities is noise in deployment terms. Unless — a state law specifically bans ALPR data retention for model training. That's where the business model actually breaks. Not the optics.
Eliza Ward: And Axon through Fusus, Motorola Solutions through Vigilant — they're watching this. But inheriting Flock's customer base means inheriting the exact same structural problem. The camera is the issue, not the logo on it.
Brian Reed: So where I land is — a hundred cancellations doesn't break Flock. Not by itself. The math is just too lopsided. But the bipartisan framing in the Washington Post, August 21st — that's the one I'd actually watch. Because a16z can wait out Berkeley canceling. They cannot wait out a Republican state legislature writing ALPR restrictions into law.
Eliza Ward: That's the signal I'd watch too. And I mean — the WaPo piece isn't just noting bipartisan *complaints*, it's framing the cameras as a bipartisan *issue*. That's a different thing. Once the ideological sorting breaks down, Garrett Langley's 'simple camera' line stops doing the work it's been doing.
Brian Reed: And that framing is under direct legal pressure now — the Institute for Justice lawsuit doesn't just challenge the policy, it challenges the premise. If a court agrees that it's functionally impossible to drive without being recorded, 'simple camera' isn't a defense anymore. It's an admission.
Eliza Ward: That case hasn't been decided. We don't know yet.
Brian Reed: No — and that's actually where it ends, isn't it? The lawsuit is live, the bipartisan test is live, and a16z is sitting on an eight-billion-dollar bet that neither of those goes badly enough to matter. I genuinely don't know who's right.
