San Diego quietly crossed a threshold this week that has long sounded like science fiction: you can now summon a car, climb into the back seat and travel across much of the city without another human being inside.
Waymo officially began welcoming public riders in San Diego on September 1, making the city one of 14 markets where the Alphabet-owned company (Google) now provides fully autonomous trips. Access is initially being rolled out gradually, but the company says tens of thousands of people in San Diego have already expressed interest and that service will continue expanding.
And Waymo may soon have company. Amazon-owned Zoox announced the same day that it plans to begin testing in San Diego, initially using conventional vehicles with humans aboard to map local streets before eventually deploying its purpose-built robotaxis. Tesla has not announced a San Diego robotaxi launch, but its unmistakable gold Cybercab prototypes have already been photographed repeatedly on local streets, while state permits allow Tesla to conduct autonomous testing in California with a safety driver.
None of this is theoretical anymore. The driverless car is here.
Not the almost-driverless car. Not a prototype circulating on a closed course somewhere in Silicon Valley. San Diegans are now being transported through Pacific Beach, downtown, Normal Heights and other neighborhoods by vehicles in which there is literally nobody sitting behind the wheel.
At the same time, ordinary consumers can already purchase something remarkably close to the experience. Tesla currently sells Full Self-Driving (Supervised), a $99-per-month system capable of navigating city streets, intersections, turns and highways while the owner remains legally and practically responsible for supervising it. Tesla itself emphasizes that the system does not make the vehicle autonomous and that the driver must remain attentive and prepared to intervene.
But Tesla's stated destination is considerably more ambitious. Elon Musk said earlier this year that he expects cars operating without human safety monitors to become widely used across the United States before the end of 2026. He has gone even further, predicting that within five to ten years roughly 90% of miles traveled could be driven by artificial intelligence, with humans manually driving cars becoming relatively unusual. Musk has made overly optimistic autonomy predictions before, and there are substantial regulatory and technological hurdles remaining, so those timelines should not be treated as guarantees. But the direction of travel is no longer particularly difficult to discern.
San Diego is now living at the beginning of it. And the question worth asking isn't simply whether driverless cars work. It is what happens if they work too well.
The immediate appeal is obvious. Autonomous vehicles don't get drunk. They don't text their friends while changing lanes. They don't become tired after twelve hours of work, develop road rage because someone cut them off or decide that a red light is merely a suggestion. For elderly people, people with disabilities and those who cannot drive, the technology could offer a degree of mobility and independence that is genuinely transformative.
There are also legitimate arguments that widespread autonomy could eventually save enormous numbers of lives. Tesla is already making aggressive safety claims about its supervised system, while Waymo has spent years presenting data intended to show that autonomous driving can reduce serious crashes compared with human drivers. Those claims deserve continued independent scrutiny, particularly because measuring autonomous-versus-human safety is far more complicated than comparing raw collision counts.
But technology rarely changes only the thing it was invented to change. The automobile didn't simply replace the horse. It reshaped cities, employment, housing, commerce, dating, tourism, policing, childhood, suburbs and the physical geography of American life. Autonomous vehicles could do the same thing in reverse.
Start with the most obvious disruption: the person who currently drives you. Uber and Lyft helped create an enormous secondary labor market in which a car became an income-producing asset. For some drivers it is a full-time occupation; for others it is the emergency financial valve that pays the rent, covers an unexpected bill or generates several hundred extra dollars each month.
Lyft says more than one million drivers earned more than $10 billion through its platform in 2025 alone. Whatever one thinks about the economics of gig work, that is an enormous amount of money currently flowing to human beings in exchange for driving other human beings around.
Now imagine the economics of rideshare when the driver disappears. A robotaxi does not need health insurance. It does not call in sick. It does not demand a raise, sleep eight hours, take a vacation, complain about gasoline prices or decide driving drunk strangers home at 1:30am is no longer worth the trouble.
Once autonomous fleets become inexpensive enough to operate at scale, human drivers will be competing against vehicles capable of working essentially around the clock. That doesn't mean every Uber and Lyft driver disappears tomorrow. It does mean an enormous category of flexible employment is staring directly at automation.
San Diego officials have already raised precisely that concern. When the Metropolitan Transit System board voted overwhelmingly earlier this year to oppose Waymo's expansion, City Councilmember Sean Elo-Rivera put the issue rather plainly: autonomous vehicles "literally replace people."
The next step is more speculative, but potentially much more profound. What happens when autonomous cars become demonstrably safer than human-driven ones?
For the moment, people tend to frame the choice as preference: Would you rather drive yourself or let the computer do it? Eventually, it may cease to be entirely a matter of preference.
Insurance is fundamentally the pricing of risk. If autonomous systems someday establish a dramatically lower accident rate than humans, insurers will have a straightforward economic reason to charge more for the riskier activity. That riskier activity could be humans actually driving the vehicle themselves.
The irony would be extraordinary. For more than a century, learning to drive represented independence. Getting a driver's license was practically a civic rite of passage. A teenager receiving the keys to a car wasn't merely acquiring transportation; they were acquiring geographic freedom.
Go wherever you want. Whenever you want. No timetable. No permission. No algorithm selecting the route.
But in an autonomous future, personally controlling a two-ton machine at 70 miles per hour may eventually look less like an ordinary activity and more like riding a horse through downtown.
Legal, perhaps. Permitted in certain places. Expensive to insure. And increasingly regarded as an eccentric hobby practiced by people who insist upon doing something machines can supposedly do more safely.
That transformation would probably happen gradually enough that few people would notice any individual step as particularly alarming. Your insurance company offers a discount if autonomous mode stays engaged. Then a larger discount. Then some roads establish autonomous-only lanes because the vehicles can travel closer together. Then an autonomous district emerges somewhere dense because eliminating unpredictable human drivers improves traffic flow.
Perhaps human-driven cars begin paying higher congestion fees. Perhaps commercial fleets refuse to insure employees driving manually except under limited circumstances. Perhaps certain freeways eventually permit autonomous vehicles at speeds or densities considered unsafe for manually controlled cars. Every individual policy could be entirely rational.
And after twenty years of rational decisions, you might discover that you still possess the legal right to drive your own automobile while having remarkably few practical opportunities to exercise it. That is where the issue becomes more philosophical than technological.
Benjamin Franklin famously wrote in 1755: "Those who would give up essential Liberty, to purchase a little temporary Safety, deserve neither Liberty nor Safety." The line originally arose from a very different dispute involving taxation, government authority and defense of Pennsylvania's frontier, so applying it literally to robotaxis would torture its historical context. But the broader tension Franklin articulated - the bargain between liberty and security - has survived because societies encounter versions of that bargain repeatedly.
Autonomous transportation may eventually present one of them. Driving is dangerous. Human beings are imperfect operators. Thousands die on American roads every year because someone was drunk, distracted, reckless, exhausted or simply made the wrong decision during one irreversible second.
If computers eventually reduce that carnage dramatically, society will understandably ask why humans should be permitted to continue creating unnecessary risk. And there is the trap.
Because the argument may be perfectly sensible. If autonomous cars become five times safer than humans, why allow people to drive through crowded downtown streets? If they're ten times safer, why let human drivers onto certain highways? If they're twenty times safer, why should an insurance company charge autonomous and manual drivers the same premium? At some point, "You don't have to use the autonomous system" can quietly become "You can technically choose not to use the autonomous system, provided you can afford the consequences."
Freedom does not always disappear because something is prohibited. Sometimes it disappears because exercising it becomes impractical.
There is another consequence that receives considerably less attention: ownership itself. When a car can arrive within three minutes whenever you need one, take you somewhere and immediately depart for another fare, owning a vehicle begins looking economically irrational for many urban residents.
That might be wonderful. No car payment. No insurance. No parking. No registration. No maintenance. Vast parking garages could become apartments, parks or businesses. City streets could reclaim enormous amounts of space currently devoted to storing machines that sit unused roughly 95% of the time.
But ownership also means control. The car in your driveway can take you somewhere even if an app goes offline. It does not need permission from a corporation's server. It does not know that your credit card expired. It cannot suddenly tell you the service area ends six blocks before your destination.
A transportation network operated largely by centralized autonomous fleets introduces a different relationship between people and movement. The machine knows where you entered. It knows where you went. It knows when you went there. It may know who accompanied you. It knows which route you took, what you paid and where you were dropped off.
And location history may ultimately be only the beginning of the privacy question, because the vehicles increasingly surrounding us don't merely know where they're going. They can see.
Waymo's robotaxis rely on an extensive array of cameras and other sensors to perceive the world around them, while cameras inside the vehicles monitor the passenger compartment. Waymo says interior cameras can be used to ensure vehicles are clean, locate lost belongings, enforce in-car rules and respond to emergencies, and acknowledges that support personnel may review recorded video and, in urgent situations, access live video from inside a vehicle. The company says it does not use facial recognition or other biometric technology to identify riders.
That makes riding in an autonomous taxi fundamentally different from getting into your own old car and driving away. A robotaxi can potentially create a digital record showing not only that you traveled from your home to a particular doctor's office, church, political meeting, bar, hotel or another person's house, but also containing video from inside the vehicle during the journey. Waymo's privacy disclosures say the company collects information including pickup and drop-off locations, trip and route information, device location and video recorded inside its vehicles.
Those records can also become relevant to law enforcement. Waymo says it may disclose information when required through valid legal process, although the company says it reviews government requests and may challenge those it considers overly broad or otherwise improper. There are perfectly legitimate reasons police might seek such evidence - investigating a kidnapping, assault, missing person or serious crash, for example - but the same technology that makes a vehicle extraordinarily observable for safety purposes also makes a person's movements potentially more reconstructable after the fact.
Consumer cars are moving in a similar technological direction, although there are important differences in what information is stored and transmitted. Teslas equipped with Full Self-Driving (Supervised) use exterior cameras to understand their surroundings and an interior cabin camera to monitor driver attentiveness. Tesla says cabin-camera images and video ordinarily remain inside the vehicle and aren't transmitted to the company unless data sharing is enabled or certain safety-related circumstances apply, while Dashcam and Sentry Mode recordings are ordinarily stored locally rather than automatically uploaded to Tesla.
Still, consider what that means at scale. A parked Tesla using Sentry Mode can use exterior cameras to detect and record activity around the vehicle, and owners can remotely view cameras through the Tesla app. Autonomous and increasingly automated vehicles therefore aren't merely putting cameras on cars so computers can drive them. They are gradually populating streets with mobile networks of sophisticated sensors capable of observing pedestrians, license plates, homes, accidents, police encounters and much of whatever else happens around them.
None of that means Waymo, Tesla or another autonomous-vehicle company is secretly watching everyone all the time, and both companies describe privacy protections governing how vehicle data is collected, stored and accessed. Much of the enormous amount of sensor information these vehicles process also never becomes personally identifiable video sitting indefinitely on a corporate server. But the technological shift changes the privacy equation nonetheless: transportation is becoming an activity capable of generating extraordinarily detailed records of where people go, when they go there, what happens around them and, in some circumstances, what happens inside the vehicle along the way.
That distinction becomes considerably more important if privately owned automobiles eventually give way to autonomous fleets. Your old car doesn't maintain a corporate account documenting every destination you visit. It doesn't associate your movements with an app, payment method and trip history, and there isn't a company sitting between you and the road that can potentially receive legal demands for information about where you traveled.
The cameras have obvious benefits. They can document crashes, protect passengers, identify vandalism, help recover lost property and provide critical evidence when something goes terribly wrong. But those same capabilities reveal another bargain accompanying the autonomous future: the safest and most convenient transportation system ever created could also become the most observable transportation system ever created.
And once again, none of this requires some sinister conspiracy. That is what makes the transition more interesting. Convenience is extraordinarily persuasive.
Benjamin Franklin famously wrote in 1755: "Those who would give up essential Liberty, to purchase a little temporary Safety, deserve neither Liberty nor Safety." The line originally arose from a very different dispute involving taxation, government authority and defense of Pennsylvania's frontier, so applying it literally to robotaxis would torture its historical context. But the broader tension Franklin articulated - the bargain between liberty and security - has survived because societies encounter versions of that bargain repeatedly.
Autonomous transportation may eventually present one of them. Driving is dangerous. Human beings are imperfect operators. Thousands die on American roads every year because someone was drunk, distracted, reckless, exhausted or simply made the wrong decision during one irreversible second.
If computers eventually reduce that carnage dramatically, society will understandably ask why humans should be permitted to continue creating unnecessary risk. And there is the trap.
Because the argument may be perfectly sensible. If autonomous cars become five times safer than humans, why allow people to drive through crowded downtown streets? If they're ten times safer, why let human drivers onto certain highways? If they're twenty times safer, why should an insurance company charge autonomous and manual drivers the same premium? At some point, "You don't have to use the autonomous system" can quietly become "You can technically choose not to use the autonomous system, provided you can afford the consequences."
Freedom does not always disappear because something is prohibited. Sometimes it disappears because exercising it becomes impractical.
There is another consequence that receives considerably less attention: ownership itself. When a car can arrive within three minutes whenever you need one, take you somewhere and immediately depart for another fare, owning a vehicle begins looking economically irrational for many urban residents.
That might be wonderful. No car payment. No insurance. No parking. No registration. No maintenance. Vast parking garages could become apartments, parks or businesses. City streets could reclaim enormous amounts of space currently devoted to storing machines that sit unused roughly 95% of the time.
But ownership also means control. The car in your driveway can take you somewhere even if an app goes offline. It does not need permission from a corporation's server. It does not know that your credit card expired. It cannot suddenly tell you the service area ends six blocks before your destination.
A transportation network operated largely by centralized autonomous fleets introduces a different relationship between people and movement. The machine knows where you entered. It knows where you went. It knows when you went there. It may know who accompanied you. It knows which route you took, what you paid and where you were dropped off.
And location history may ultimately be only the beginning of the privacy question, because the vehicles increasingly surrounding us don't merely know where they're going. They can see.
Waymo's robotaxis rely on an extensive array of cameras and other sensors to perceive the world around them, while cameras inside the vehicles monitor the passenger compartment. Waymo says interior cameras can be used to ensure vehicles are clean, locate lost belongings, enforce in-car rules and respond to emergencies, and acknowledges that support personnel may review recorded video and, in urgent situations, access live video from inside a vehicle. The company says it does not use facial recognition or other biometric technology to identify riders.
That makes riding in an autonomous taxi fundamentally different from getting into your own old car and driving away. A robotaxi can potentially create a digital record showing not only that you traveled from your home to a particular doctor's office, church, political meeting, bar, hotel or another person's house, but also containing video from inside the vehicle during the journey. Waymo's privacy disclosures say the company collects information including pickup and drop-off locations, trip and route information, device location and video recorded inside its vehicles.
Those records can also become relevant to law enforcement. Waymo says it may disclose information when required through valid legal process, although the company says it reviews government requests and may challenge those it considers overly broad or otherwise improper. There are perfectly legitimate reasons police might seek such evidence - investigating a kidnapping, assault, missing person or serious crash, for example - but the same technology that makes a vehicle extraordinarily observable for safety purposes also makes a person's movements potentially more reconstructable after the fact.
Consumer cars are moving in a similar technological direction, although there are important differences in what information is stored and transmitted. Teslas equipped with Full Self-Driving (Supervised) use exterior cameras to understand their surroundings and an interior cabin camera to monitor driver attentiveness. Tesla says cabin-camera images and video ordinarily remain inside the vehicle and aren't transmitted to the company unless data sharing is enabled or certain safety-related circumstances apply, while Dashcam and Sentry Mode recordings are ordinarily stored locally rather than automatically uploaded to Tesla.
Still, consider what that means at scale. A parked Tesla using Sentry Mode can use exterior cameras to detect and record activity around the vehicle, and owners can remotely view cameras through the Tesla app. Autonomous and increasingly automated vehicles therefore aren't merely putting cameras on cars so computers can drive them. They are gradually populating streets with mobile networks of sophisticated sensors capable of observing pedestrians, license plates, homes, accidents, police encounters and much of whatever else happens around them.
None of that means Waymo, Tesla or another autonomous-vehicle company is secretly watching everyone all the time, and both companies describe privacy protections governing how vehicle data is collected, stored and accessed. Much of the enormous amount of sensor information these vehicles process also never becomes personally identifiable video sitting indefinitely on a corporate server. But the technological shift changes the privacy equation nonetheless: transportation is becoming an activity capable of generating extraordinarily detailed records of where people go, when they go there, what happens around them and, in some circumstances, what happens inside the vehicle along the way.
That distinction becomes considerably more important if privately owned automobiles eventually give way to autonomous fleets. Your old car doesn't maintain a corporate account documenting every destination you visit. It doesn't associate your movements with an app, payment method and trip history, and there isn't a company sitting between you and the road that can potentially receive legal demands for information about where you traveled.
The cameras have obvious benefits. They can document crashes, protect passengers, identify vandalism, help recover lost property and provide critical evidence when something goes terribly wrong. But those same capabilities reveal another bargain accompanying the autonomous future: the safest and most convenient transportation system ever created could also become the most observable transportation system ever created.
And once again, none of this requires some sinister conspiracy. That is what makes the transition more interesting. Convenience is extraordinarily persuasive.
People did not surrender personal information to smartphones because governments forced them to. We willingly carried devices capable of tracking our locations because they also gave us maps, cameras, music, banking, dating, restaurant reservations and practically the sum of accumulated human knowledge. The bargain was too useful to refuse.
Autonomous transportation could be exactly the same. The Waymo arrives faster than Uber. It is cheaper. It doesn't expect a tip. You can work during the ride. Your teenager can get home without borrowing the car. Your 85-year-old parent can retain independence. Crashes decline. Traffic improves. Insurance gets cheaper if you stop driving. And eventually you wonder why you still own a steering wheel.
There is nothing inherently dystopian about that future. In many respects it could be better than the present one. But progress deserves examination precisely when its benefits are compelling.
Waymo's arrival in San Diego this week is easy to treat as another quirky technology story: look, there goes the little car with nobody driving. It is more significant than that. A machine has entered the transportation economy capable of performing one of the most ordinary human tasks without a human being.
Autonomous transportation could be exactly the same. The Waymo arrives faster than Uber. It is cheaper. It doesn't expect a tip. You can work during the ride. Your teenager can get home without borrowing the car. Your 85-year-old parent can retain independence. Crashes decline. Traffic improves. Insurance gets cheaper if you stop driving. And eventually you wonder why you still own a steering wheel.
There is nothing inherently dystopian about that future. In many respects it could be better than the present one. But progress deserves examination precisely when its benefits are compelling.
Waymo's arrival in San Diego this week is easy to treat as another quirky technology story: look, there goes the little car with nobody driving. It is more significant than that. A machine has entered the transportation economy capable of performing one of the most ordinary human tasks without a human being.
Zoox is coming behind it, brought to you by Amazon. Tesla appears to be circling San Diego as well, while simultaneously selling technology to ordinary vehicle owners that already handles most of the mechanical act of driving, just "under supervision." The primitive versions are arriving simultaneously from several other directions.
Twenty years from now, people may look back at 2026 as the period when autonomous cars finally became mundane. Children born today may find the idea that millions of distracted, emotional, sleep-deprived humans once independently piloted vehicles past one another at freeway speeds completely insane. They may be right.
They may also inherit a world in which driving a car yourself is expensive, geographically restricted, socially frowned upon or simply unnecessary. Perhaps that is progress. Perhaps the roads become dramatically safer, transportation becomes cheaper and mobility becomes available to millions of people currently excluded from it. They may also find it completely ordinary that nearly every trip they take occurs inside and around vehicles capable of seeing, recording and remembering far more about their movements than the automobile ever could before.
But it is worth recognizing the bargain while we are still early enough to see it being made. Because the future rarely arrives with a ceremony announcing which freedoms are about to become obsolete.
Sometimes it simply pulls up to the curb. And there is nobody behind the wheel.
Originally published on September 2, 2026.
