The Waymos Will Kill Us All

Waymo vehicles committed twenty-three criminal violations against Austin school buses in a single school year. Criminal — not regulatory, not procedural: criminal. Under Texas Transportation Code §545.066, passing a school bus with active stop-arm signals is a criminal offense. Twenty-three times, Waymo's fleet did exactly that.

Austin Independent School District (ISD) documented it all. They counted. They compiled the records. Then they did what you'd expect a school district to do when a company is committing serial criminal violations near children: they asked the company to stop operating during student pickup and dropoff hours. A reasonable request. An obvious request. The minimum ask.

Waymo refused.

Austin Police Chief Wayne Sneed (Bloomberg, January 2026), on the record: "They did not agree with our risk assessment and respectfully declined to stop operating."

Waymo had already filed a voluntary software recall with NHTSA in December 2025, claiming the problem was fixed. The January 12, 2026 violation happened after the claimed fix — when a Waymo vehicle's remote operator, asked whether a school bus ahead had active signals, said "No" and the vehicle proceeded through. The March 25, 2026 violation happened after that.

Council Member Paige Ellis (Austin Current, April 2026): "The question is not if this is going to turn into a deadly situation but when." Waymo's response: a press release about safety being at the core of everything they do.

Here is the thing that should make you stop and sit with it for a moment.

If a human rideshare driver passed a stopped school bus three times, they would lose their license. That is the deal. The accountability rail for individual drivers is built around removing the violator from the road. Serial violation → disqualification. The logic is simple. The enforcement is real. It works because the violator is an individual who can be identified, sanctioned, and removed.

Waymo's entire fleet operates with the same brain — the same software, the same decision-making architecture, the same routing logic. When the brain decides to pass a school bus, every vehicle running that brain makes the same decision. That brain has now committed twenty-three criminal violations in Austin. It has done this repeatedly, across multiple calendar months, after a voluntary recall that Waymo itself filed acknowledging the exact problem. The individual-driver accountability framework says: remove the violator. But you cannot remove "Waymo." You can't revoke its license. You can only try to shut down a multi-billion-dollar corporate operation with armies of lobbyists, state preemption laws designed precisely to prevent local governments from acting, and political donations flowing to officials in both parties.

No regulator has that political appetite. So the same brain keeps driving.

This is not a Waymo-specific failure. This is what happens structurally when the individual-driver accountability framework encounters a corporate fleet that runs a unified decision-making system. The framework was designed for one driver, one violation, one sanction. It has no mechanism for one brain, fleet-scale violation, and zero corporate accountability. The gap between those two things is not a bug in how Waymo was implemented. It is the architecture. It is the point.

The question worth asking — and asking loudly, before this is everywhere — is who benefits from an accountability vacuum that large.

Waymo vehicles are not cars with software. They are rolling computers with cellular uplinks, cloud-managed control planes, and remote operators who can redirect a stopped vehicle from a call center. That architecture is the product. It is also the attack surface.

The same remote-access infrastructure that lets a Waymo operator override a vehicle from headquarters is the same infrastructure that lets a sufficiently motivated hostile actor do exactly the same thing. Cellular uplinks are attack surfaces. Cloud control planes are attack surfaces. The corporation behind the fleet is an attack surface — insider threats, ransomware, nation-state intrusion, disgruntled former engineers with valid credentials and no severance package. The threat model is not limited to "what does Waymo's corporate management decide to do." It is "what happens when Waymo loses control of its fleet, even briefly, because someone else took it."

Critical-infrastructure ransomware is not a hypothetical. In May 2021, DarkSide ransomware hit Colonial Pipeline — one company, one pipeline — and shut down fuel supply to the East Coast for six days. The attackers got in through a single compromised password on a legacy VPN account. President Biden declared a state of emergency for 18 eastern states. The FBI, CISA, and NSA jointly attributed the attack and issued a national critical-infrastructure security advisory. The company paid $4.4 million in ransom. Congress held hearings. The CEO apologized. The pipeline moved fuel.

Waymo moves people. The attack surface is the same architecture: a centralized corporate control plane managing critical transportation infrastructure across multiple cities, connected to the internet, operated by humans who can be phished, coerced, or bought. The documented record of critical-infrastructure ransomware says this is not a question of whether. It is a question of when — and of who controls the fleet when it happens.

In 2015, security researchers Charlie Miller and Chris Valasek remotely killed a Jeep Cherokee on a highway — cut the engine, disabled the brakes, blasted the air conditioning — from a laptop ten miles away. The story ran in Wired. Chrysler issued a recall of 1.4 million vehicles. Congress introduced legislation. It was treated as a crisis.

The blast radius: one car.

That is the ceiling with private-ownership autonomy. Individual vehicles require individual compromises — but not because private cars run independent software. They don't. Every Tesla runs the same firmware pushed from the same central OTA server; a single vulnerability can hit an entire private OEM fleet, as both the 2015 Jeep hack and the 2023 2M-vehicle Tesla recall (NHTSA 23V838) demonstrate. What private ownership actually limits is something different: there is no continuously-active real-time operator control plane; no routing and dispatch layer as a separate attack surface; and the vehicles are parked most of the time, not continuously deployed in public space. The blast radius ceiling stays low because the attack surfaces are fewer — not because the software is independent.

Corporate-fleet centralization is one target with thousands-of-times-the-impact per compromise. When Waymo's central control plane is taken — ransomware, nation-state intrusion, insider threat, corporate capitulation to a government demand — the blast radius is every vehicle in the fleet, simultaneously, in every city where Waymo operates. Not one car. Not one city. San Francisco, Los Angeles, Phoenix, Austin, Atlanta, Miami — all at once. An attacker who wants to paralyze urban mobility across the United States does not need to compromise 400,000 cars. They need to compromise one control plane. The cost per vehicle affected approaches zero. The damage ceiling is every car.

The autonomous vehicle industry's safety case is about reducing accidents per mile. It has no answer for what happens when the centralized brain that reduces accidents per mile becomes the single point of failure in critical urban transportation infrastructure. One successful intrusion. Six cities. At once.

A gunman opened fire at Buford's Backyard Beer Garden on West 6th Street. Three people killed. Fifteen injured. Paramedics responded. A Waymo vehicle sat stopped in the road and would not move. The ambulance rerouted around it — approximately two minutes before an Austin police officer drove the vehicle manually to a nearby parking garage.

Two minutes. Three dead. The paramedics rerouted.

The NTSB was already investigating Waymo's school bus violations — not Waymo's first federal probe in Austin. Now this.

Waymo's official response: "Safety is at the core of everything we do."

When the same brain that refused to stop during school pickup hours encounters an ambulance at a mass shooting, it does what it was designed to do. It parks. It waits for instruction. It does not recognize that this is the moment the accountability vacuum starts to matter in ways that are counted in bodies. This is not a hypothetical about what could go wrong. This is what did go wrong. The documented record is the floor.

Each Waymo vehicle carries 29 cameras, lidar, and radar — full 360° coverage, running continuously, recording everything within sensor range. Microphones are present. The company claims they are used only for siren detection and rider-initiated calls. That claim is not verifiable. The architecture supports broader audio capture. You are asked to trust the company's word on this. The company that just refused to stop passing school buses.

As of early 2026, Waymo completes over 500,000 rides per week across San Francisco, Los Angeles, Phoenix, Austin, Atlanta, and Miami. That is over 500,000 trips per week of a moving 29-camera sensor array recording every street it passes — protests, immigration checkpoints, abortion clinics, union halls, private residences, public gatherings. Everything.

Law enforcement is already using the footage. Search warrants have been served to Waymo and Cruise across San Francisco and Maricopa County. In April 2025, LAPD obtained a warrant for Waymo footage and published it on YouTube to solicit tips. SFPD has requested footage for hit-and-runs and other crimes.

WIRED asked Waymo: how long is footage retained? Have recordings been provided to U.S. federal law enforcement — FBI, ICE, DHS, NSA? What are the full camera specs? Waymo declined to answer any of those questions. The company publishes a transparency report for local law enforcement requests only. Federal agency access: undisclosed. Officially: they won't say.

Dave Maass of the Electronic Frontier Foundation: "What starts as investigating hit-and-runs could easily expand to tracking protesters or monitoring everyday activities." Yes. That is the concern. That is also, given who built this company and for what purposes, not an accident.

Waymo did not ask Austin if it could operate there. There was no referendum. No city council vote. No public process. State-level approval — obtained from legislators who had already stripped cities of any regulatory role — was the only mechanism. And the stripping happened years before Waymo showed up.

Texas banned city-level regulation of autonomous vehicles in 2017. The industry lobbied for this. They wrote model legislation and handed it to state legislators. Then they waited for the product to be ready. When it was, they deployed into cities that had been legally disarmed in advance. The timing is not coincidental. The legal preemption was the plan.

When twenty-three school bus violations accumulated and Austin ISD asked for a suspension of operations, the city had no legal authority to require it. Waymo could — and did — simply say no. When an ambulance got blocked at a shooting, the city could issue a statement. That's the available tool. A statement. Against a company with a market cap larger than most national economies.

Texas's new AV oversight law, giving the state meaningful enforcement authority, takes effect at the end of May 2026. After the twenty-three violations. After the NTSB investigation. After the voluntary recall that didn't work. After the blocked ambulance. After three people were already dead.

This is the pattern across more than 35 states. Most AV laws include municipal preemption clauses. Georgia law declares the General Assembly “fully occupies and preempts the entire field” of AV regulation. Cities are barred by statute. Not one city voted for this. Not one.

Missy Cummings, director, Mason Autonomy & Robotics Center, George Mason University (Austin Current, April 2026): "Austin is being treated as a lab experiment that they didn't sign up for." That is correct. It is also, from the perspective of the company that lobbied for the preemption and then deployed into the preempted city, precisely as intended.

The same brain that decided school district authority didn't apply to it — that same architecture, at scale, across every American city — is worth thinking about carefully. Not as a hypothetical. As a logical extension of what is already documented.

If the brain doesn't recognize school-district authority over its operations, why would it recognize anyone else's? If it has already refused one formal public safety request from a city government, what is the mechanism by which it responds to the next one differently? The voluntary recall that "fixed" the school bus problem didn't fix it. The assurances continue. The violations continue. The same brain keeps driving.

Fleet ownership is mobility control

When Alphabet owns the fleet, Alphabet decides who moves, where, and on what terms. Today: surge pricing during sports events. At dominance: surge pricing during protests, evacuations, election days. The right to set the price of a ride to the hospital, the picket line, or the polling place — concentrated in a single corporate shareholder vote. No city council. No public utility commission. One board.

The cutoff capability

Service-refusal is an established Big Tech tool. Banks freeze accounts. Platforms ban users. Cloudflare drops sites. Apply that to physical mobility. Waymo's terms of service already permit ride refusal for "policy violations." At dominance, that is de-platforming applied to the right to move through a city. A worker on a picket line. A journalist covering a story Alphabet's advertisers dislike. A city that voted against the company's preferred zoning. Once the alternative transportation infrastructure is gone, the fleet operator decides who rides and who walks. The same brain decides. At scale.

Surveillance state extension

29 cameras per vehicle today, deployed across U.S. cities. At dominance: a continuous corporate sensor network across every American street. Every protest. Every immigration checkpoint. Every union meeting. Every abortion clinic visit. Every Pride march. Recorded by infrastructure with no warrant requirement for federal access — and with a company that already won't say which federal agencies have received footage. What that looks like when most street-level recording infrastructure runs through one Alphabet subsidiary is not a conspiracy theory. It is a straight-line projection from the architecture that already exists.

Cities redesigned for the fleet

Once AVs are dominant, cities get redesigned for AVs: roads optimized for sensor-friendly layouts rather than pedestrians, parking eliminated in favor of AV-staging zones, public transit defunded because "we have AVs now." Each step compounds. Alternatives weaken. Dependency deepens. The fleet operator acquires more leverage over the cities that depend on it. The pattern is familiar — Walmart and small-town retail, Amazon and independent booksellers, Uber and the taxi industry. Once the alternative infrastructure is gone, the price is whatever the operator wants.

None of this has happened yet

These are capabilities, not confirmed operations. The school bus violations are documented. The ambulance block is documented. The surveillance architecture is documented. The federal agency opacity is documented. The rest follows from the logic of what has already been built, deployed, and defended. You are watching a company establish that it does not have to answer to school districts, city councils, or safety agencies when it disagrees with their risk assessments. The question is what it does with that precedent once it has the whole road.

According to analysis by Gadallon (Gadallon Substack; independent verification pending), from July 2024 to July 2025, median rideshare pay per trip increased 3.4% nationally. In every city where Waymo operates at scale, hourly pay moved in the opposite direction:

• San Francisco: -6.9% hourly year-over-year
• Austin: -5.3% hourly year-over-year
• Phoenix: down ~4–5% hourly year-over-year
• Los Angeles: roughly flat (early-stage Waymo presence)

Over 500,000 rides per week means over 500,000 trips human drivers did not take. Each Waymo ride is a trip extracted from the labor pool of gig workers who depend on rideshare income. This is not displacement that hasn't happened yet. It is happening now, in the wage data, city by city. Correlation with Waymo rollout does not isolate Waymo as sole cause — but the direction is consistent across every market where Waymo operates at scale.

At least 4 million driving jobs — rideshare, taxi, delivery, trucking — are at risk in the long-run transition to autonomous vehicles (Goldman Sachs Global Investment Research, 2023; consistent with McKinsey Global Institute projections). The capital that owns the fleet captures the value those workers were previously paid. This is described in press releases as innovation. Structurally, it is the extraction of labor income into shareholder return.

The workers whose livelihoods are being extracted by Waymo's fleet were not consulted when the system was built, were not compensated when their routes became training ground for the technology's market development, and were not represented in the lobbying that preempted local oversight before the product existed. The capital that owns the fleet now captures the value those workers were previously paid. That is the transaction. That is the whole transaction.

Waymo is an Alphabet subsidiary. Alphabet’s foundational business is surveillance — ad targeting built on tracking what users do online, where they go, what they search, who they know. The engineering culture that built Waymo is the same culture that built Google’s ad-tracking stack. The 29 cameras per vehicle aren’t a coincidence; they’re consistent with the parent company’s lifelong design philosophy.

Brin and Page's foundational research at Stanford was conducted under a grant program co-sponsored by the CIA and NSA — the Massive Digital Data Systems program, jointly sponsored with the Director of Central Intelligence's Community Management Staff. The connection is documented in a footnote of the 1998 paper that introduced PageRank. The funding went to Stanford, not to Google as a company; Google was separately venture-funded. The connection is not investment. It is that the core algorithmic work grew in an environment where intelligence-agency money was at the table.

In 2003, the CIA's venture capital arm, In-Q-Tel, invested in Keyhole, Inc. — the satellite mapping company Google acquired in 2004. Keyhole became Google Earth. By 2003, Google had a special contract with the CIA for its Intelink Management Office.

Waymo did not inherit a direct CIA investment. It inherited a corporate culture built by people who never treated government surveillance access as a design problem to be solved — because from the beginning, it wasn't. The company that will not say which federal agencies have the footage from over 500,000 weekly rides is the successor to a company that was seed-funded by the agencies that would like that footage. These things are connected. You are not required to call it a conspiracy to notice that the connections exist.

The railway barons of the 19th century built rail on massive public subsidy — federal land grants, government bonds, surveyed public land handed over at below-market rates. Once the track was laid, they leveraged it: setting freight rates that broke farmers, controlling state legislatures, bribing federal officials, breaking strikes with private armies and Pinkerton detectives.

The pattern is: build infrastructure the public cannot do without, capture the regulatory apparatus that might constrain you, and then extract. The extraction phase always looks like the inevitable consequence of progress. The farmers and small merchants who got broken by freight rates were told this was the price of modernization. The gig workers watching their wages fall are being told the same thing now.

Waymo is on the same arc — with additional capabilities the railway barons could not have imagined. The fleet records everyone it passes and hands those recordings to agencies that file the right form. The preemption laws were written before the product arrived. The regulatory capture is already complete in most states. And the company has already established, in the documented record, that it does not have to stop when a school district asks it to.

The railway monopolies were eventually broken — by the Interstate Commerce Commission, by the Sherman Act, by decades of political organizing by the people they were breaking. The question is whether that happens before or after the dominance phase. Right now, in the convenience phase, with only 500,000 rides per week and only ten cities and only twenty-three criminal violations and only one blocked ambulance — right now is when that question can still be answered before the answer is obvious.

Take everything documented in this thread. The centralized control — one brain in every vehicle, across every city, updated simultaneously by one corporate operator. The established pattern of refusing public authority — a school district's formal request, declined. The 29-camera surveillance suite recording every street, every gathering, every face the fleet passes. The existing federal law enforcement cooperation, with the scope of that cooperation undisclosed. The intelligence-community lineage of the company that built it. The legal preemption of any local government that might object.

Now ask: what would it actually take to convert this fleet into law enforcement infrastructure?

Almost nothing. A government contract. Payload mounts. The vehicles are already deployed in every major American city. The operators already control every car remotely. The cameras already record everything in their path. The brain is centralized — one instruction from headquarters reaches the entire fleet simultaneously. Add tear-gas dispensers. Add loudspeakers for disperse announcements. Add the legal apparatus to declare curfews enforced by autonomous vehicles. Add the political apparatus to declare a national security emergency and direct the fleet to enforce it.

None of this requires building new infrastructure. The infrastructure exists. The vehicles are in position. The control plane is operational. The corporate operator has already demonstrated — in the documented record, in Austin, Texas, about a specific public-safety request from an elected school district — that it will not comply with demands from local governments when it respectfully disagrees with their risk assessment. The question of whether it would comply with demands from a federal government declaring a different kind of emergency has a different answer. The federal agency cooperation is already documented. The corporate culture was built by people who never treated that relationship as a problem.

This is what a private robot army looks like. Not in the speculative future. Today. In every city where Waymo operates. Lightly disguised as a ride-share app. The conspiracy theory version of this argument adds sinister intent. The documented version requires only the architecture that already exists and the precedents already set.