A man trapped in a self-driving car is now the most searched robotaxi story of the week, and the headline is built around a high-tech search for his home. The Anniston Star, 28/22 News and The Jerusalem Post are the outlets named as carrying that story. The full text of those reports was not available for this analysis, so this article does not restate their details as fact. What it can do is lay out the two trapped-rider incidents that reports have documented in recent weeks and explain why the hardware behind them keeps failing in the same way.
The first involves Mike Johns, a Los Angeles tech entrepreneur who was allegedly stuck in a Waymo on the way to the Phoenix airport. The second involves David Moss, a Texas man allegedly held for 14 minutes inside a Tesla Cybercab on October 5, with about 7 minutes needed just to restart FSD. Both accounts remain unconfirmed by the companies, and we flag every claim below accordingly. For context on how fast the Tesla side of this market is growing, see our report on the Cybercab fleet growing 8x in 30 days.
This piece sits in our hardware coverage because the story is at its core about sensors, compute and the physical controls inside a car with no driver. A trapped rider is a hardware and software failure that a passenger can see, feel and time with a phone stopwatch.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What the Reports Say About the Man Trapped in a Self-Driving Car
The headline reads “High-tech search for home by man trapped in self-driving car.” Taken at face value, it describes a rider who could not simply tell the vehicle where to go, and a technology-assisted effort to locate or reach his home. Beyond that framing, we have no verified detail. We do not know whether the car, the rider or a remote operator ran the search. We also do not know the vehicle brand, the city, the length of the ordeal or whether the company commented.
That gap matters because several different robotaxi incidents are circulating under similar search terms. A reader who types “man trapped in self-driving car” will find the Waymo airport case, the Cybercab case and older incidents involving other riders. Treating them as one event would be a mistake. Our approach here is to keep each incident separate, attribute each claim to the outlet or party that made it, and mark anything the sources do not support.
Two outlets we do cite for supporting detail are Tesla North, which reported on the October 5 Cybercab ride, and Gadget Review, which described a separate Austin hotel parking lot loop. Neither report should be read as an official finding. They are press accounts of rider experiences, and the companies involved have not, in the sources we reviewed, confirmed the timelines.
What We Could and Could Not Confirm
Honest reporting on this topic starts with a short list of what is missing. The sources available for this article do not establish an exact official product name tied to the headline event. They do not contain a company statement for it, a confirmed price, a technical specification or an on-record quote from anyone involved. They also show that the results refer to multiple separate incidents involving Waymo robotaxis and a Tesla Cybercab, so the headline cannot be reliably pinned to one of them.
What the sources do support is narrower. A result identifies Mike Johns as a Los Angeles-based tech entrepreneur who was allegedly trapped in a Waymo while traveling toward the Phoenix airport. The same result says the vehicle drove in circles for several minutes and nearly caused him to miss his flight. Another result identifies David Moss as a Texas man allegedly trapped for 14 minutes in a Tesla Cybercab on October 5. Every one of those statements carries the word “allegedly” for a reason.
We chose not to quote anyone directly in this article. No verified quote set was available to us for this story, and a reworded quote placed inside quotation marks would be a fabrication. Where a company position is described, we paraphrase it and say it was reported.
The Waymo Case: Mike Johns and the Phoenix Airport Loop
According to the account tied to Mike Johns, the Waymo he rode toward the Phoenix airport drove in circles for several minutes. He was on a deadline, so the delay nearly cost him his flight. The same account says Waymo responded that its vehicles have a “pull over” button available to riders and that a software update addressed the issue.
Read carefully, that response contains two separate admissions of design shape, even if not of fault. First, the company points to a rider-facing control as the intended escape hatch. Second, it says software was changed afterward, which implies the looping behavior was reproducible enough to patch. Neither point tells us how many rides were affected, how long the issue existed or whether the button was visible and obvious to a stressed passenger.
Waymo is not a newcomer to scrutiny. Our earlier piece on a Tesla robotaxi dodging a Waymo and an Avride vehicle shows how closely rival fleets now share the same streets in the same cities. Separately, the BBC reported that Waymo paused service in five U.S. cities after vehicles reportedly entered floodwater, following a San Antonio incident in which an empty robotaxi was swept into a creek. Yahoo News also reported that multiple Waymo vehicles lost power during Fourth of July crowds and had to be towed one at a time. None of those events involve the Phoenix rider, but together they show a pattern: when a robotaxi leaves its comfort zone, the rider has little recourse.
The Tesla Cybercab Case: David Moss and 14 Minutes
The Texas account is the more detailed of the two. David Moss was allegedly trapped inside a Tesla Cybercab for 14 minutes on October 5. Roughly 7 minutes of that window reportedly went to restarting FSD, Tesla’s driving software. Tesla North, which covered the ride, reported that the car spent about seven minutes looping before Tesla support remotely stopped it, and that support then needed another seven minutes to restart the software before the trip could continue.
Timeline of the reported Cybercab ride
Laid out in order, the account runs as follows. The vehicle begins a trip and fails to complete it. It circles for around seven minutes, according to Tesla North. Tesla support intervenes remotely and halts the car. Support then spends about seven minutes restarting FSD. The total comes to 14 minutes from the first loop to a working vehicle. If the account is accurate, the rider had no way to speed up either phase.
A second Austin report, from Gadget Review, describes a Cybercab that looped around a hotel parking lot four times, failed to identify an open lane near construction and told the passenger he had arrived at the starting point. Gadget Review’s account involves a different passenger and a different location, so we treat it as supporting evidence of a pattern rather than a duplicate of the Moss ride.
Why the vehicle design raises the stakes
The Cybercab is built around the absence of a human driver. As we covered in our look at the Cybercab’s 2 seats and 0 steering wheel, there is no manual fallback inside the cabin. In a conventional car, a frustrated passenger can ask a driver to stop or can take over. In a vehicle without controls, the only exits are a software command and a door. That is why a 14-minute stall feels so much longer to the person sitting in it.
Reported Incidents Side by Side
The table below puts the incidents named in this article in one place. The final column shows how each is sourced, because the strength of evidence varies widely. Treat any row marked unconfirmed as an allegation.
| Incident | Vehicle | Reported detail | Source status |
|---|---|---|---|
| Phoenix airport trip | Waymo | Mike Johns allegedly trapped, car drove in circles for several minutes, flight nearly missed | Unconfirmed |
| Texas ride, Oct. 5 | Tesla Cybercab | David Moss allegedly trapped 14 minutes, about 7 minutes to restart FSD | Unconfirmed, covered by Tesla North |
| Austin hotel lot | Tesla Cybercab | Four loops, missed open lane near construction, told passenger he had arrived | Reported by Gadget Review |
| San Francisco, May | Waymo | Passenger trapped in stationary car after windshield was cracked by two men | Reported by Yahoo News |
| Five U.S. cities | Waymo | Service paused after vehicles reportedly entered floodwater, no rider involved | Reported by BBC News |
| “Search for home” story | Not identified | Headline only, details not verified | Named outlets: Anniston Star, 28/22 News, Jerusalem Post |
Why Robotaxis Get Stuck: The Hardware and Software Stack
A looping robotaxi is rarely a single bug. It usually reflects a mismatch between what the cameras, radar or lidar report and what the planning software is allowed to do. When the planner cannot find a legal, safe path to the destination, many systems default to the most conservative action, which can mean circling a block or a parking lot until something changes.
Sensors and compute
The Gadget Review account of the hotel lot is a textbook case. Construction altered the layout, the vehicle could not identify an open lane, and the software fell back to a loop. Nothing in the report suggests a broken sensor. It points instead to a map or perception mismatch that the planner could not resolve. Map-dependent systems handle the common case well and struggle at the edges, such as temporary barriers, private lots and drop-off zones.
Tesla has said it is moving toward its own mapping layer, a shift we covered in Tesla Builds Own FSD Maps, Starting With v15. If that work lands, it could reduce the kind of private-lot confusion described above. Until then, the edge cases fall to remote support.
The compute side matters too. A software restart that takes about 7 minutes, as reported for the Cybercab, suggests the recovery path is slow compared with the failure itself. A rider who has been circling for seven minutes then waits another seven for a reboot. Shortening that second number is a pure engineering problem, and it is one of the few that a rider would immediately notice.
The High-Tech Search for Home: How Riders Find a Way Out
The headline phrase “high-tech search” is the part we cannot verify, so we will describe what such a search generally involves rather than claim it happened. A robotaxi rider has three layers of help. The first is the in-car interface, which usually offers a stop or pull-over request. The second is the app, which can show location and let the rider message support. The third is a remote operator who can see the vehicle’s state and send commands.
If a rider is disoriented or the destination was entered incorrectly, any of those layers could be used to locate a home address. Whether the man in the headline relied on one of them, on a phone map or on emergency responders is unknown. Yahoo News reported that Waymo has said passengers generally cannot independently obtain interior or exterior vehicle footage, and that law enforcement may need a search warrant. That detail suggests the rider-facing evidence trail is thin, which complicates any after-the-fact investigation.
For readers who want a practical takeaway, the answer is simple. Know where the stop control is before the car moves, keep your phone charged, and share your trip with a contact. We expand on that in the safety checklist later in this piece.
Regulators Are Watching Robotaxi Reliability
Trapped-rider stories land in an environment where federal regulators are already looking hard at autonomous fleets. We reported that NHTSA is probing 1,000 Cybercabs as Tesla eyes sales, and the agency’s AQ26002 audit of Cybercab deliveries is covered in a separate piece on the Cybercab delivery push and the audit.
Neither of those actions is described in our sources as being about the Moss ride. Still, a 14-minute stall with a 7-minute software restart is the sort of event a regulator would want to see logged and counted. If operators report these cases consistently, a pattern becomes visible. If they do not, the public learns about them one social post at a time.
The core regulatory question is not whether a robotaxi will ever stall. It will. The question is whether the stall is safe, short and recoverable by the person inside. Today, the evidence for that is mostly anecdotal.
Market Impact: Trust, Fleet Growth and Insurance
Robotaxi economics depend on rider trust. A passenger who spends 14 minutes in a looping car is unlikely to book again without a reason to believe it was a one-off. That is a direct revenue risk for any operator trying to scale a service from a handful of cities to many.
Tesla is pushing in the opposite direction, toward faster expansion. Our report on the Q3 call confirming Cybercab sales by year-end shows the company wants vehicles in more hands soon. Faster rollout raises the number of rides, and with more rides comes a higher absolute count of edge-case failures, even if the failure rate stays flat.
Insurers and municipal partners will also read these stories. A service that cannot show rider-exit procedures will face tougher questions on liability and permits. Operators that publish clear numbers on stall duration and remote response time will have an advantage. At present, none of the sources we reviewed provide those numbers for either company.
Competitive Comparison: Waymo vs Tesla on Rider Controls
The table below compares what has been reported for each operator. It is not a spec sheet, because neither company’s sources give a verified technical comparison. Cells that read “Not confirmed” mean the research did not establish the fact.
| Question | Waymo (Phoenix account) | Tesla Cybercab (Texas account) |
|---|---|---|
| Reported duration | Several minutes of circling | 14 minutes total |
| Rider-facing control | Waymo reportedly points to a “pull over” button | Not confirmed |
| Remote intervention | Not confirmed | Tesla support reportedly stopped the car remotely (Tesla North) |
| Software recovery time | Not confirmed | About 7 minutes to restart FSD |
| Fix reported | Software update reportedly addressed the issue | Not confirmed |
| Vehicle layout | Not covered in sources | 2 seats and no steering wheel, per our earlier report |
| Company statement | Reported, not verified on the record | None found in sources |
The takeaway from the table is about communication as much as engineering. Waymo reportedly gave a specific answer, a button plus an update. Tesla, in the sources we reviewed, did not. That difference may shape how riders judge each company the next time a car circles.
Historical Context: From Safety Drivers to Empty Cars
For most of the last decade, autonomous test vehicles carried a human who could grab the wheel. The shift to driverless rides removed that person and replaced them with remote support staff and software recovery routines. The Aurora trucking pilot we covered, driverless trucks logging 1,200 miles as an observer returns, shows that driverless operators still weigh whether to keep a human on board.
That history explains why a trapped-rider story hits a nerve. Every earlier generation of testing had a human backstop. The current generation does not, at least not inside the passenger compartment. The backstop moved to a call center, and the quality of that call center now defines the rider experience during a failure.
Older incidents show the pattern is not new. The AI Incident Database records a September 30, 2024 case in which a passenger was inside a Waymo blocked by two men in San Francisco, and it notes that Waymo Rider Support intervened. It also lists a December 8, 2025 report of a man found in a Los Angeles Waymo’s trunk. Those cases differ from a software loop, but they share a feature: the rider relied on remote help to resolve a situation the car could not handle alone.
Five Predictions for Trapped-Rider Incidents
The following are our analysis, not reported facts. We base them on the incidents above and on how regulators and operators have reacted to earlier problems.
- Operators will publish rider-exit procedures. Expect a short, standard guide on stop controls and door release to appear in apps and on vehicle screens within the next year.
- Stall duration becomes a tracked metric. Regulators already examining Cybercab fleets are likely to ask for time-to-recovery data, not just crash counts.
- Remote support gets faster or gets scrutinized. A 7-minute software restart is hard to defend once it is public, so recovery routines will be a priority fix.
- Hardware gets a manual override question. Vehicles with no steering wheel will face pressure to add a clearly marked physical or on-screen stop that works even when the main software hangs.
- Search interest keeps spiking on each new account. Stories like the Moss and Johns rides spread quickly, so expect more rider-submitted videos and more press coverage as fleets grow.
What Riders Can Do Today
Nothing here replaces official guidance from Waymo or Tesla, so check each app’s help page. With that caveat, a few habits make a bad ride shorter. Find the stop or pull-over control before the vehicle starts moving. Share your live trip with a friend or family member. Keep a charged phone in hand, and note the time the problem starts so you can report an accurate duration afterward.
If the car loops, contact in-app support first, then call emergency services if you feel unsafe. Do not try to exit a moving vehicle. Save screenshots of the trip map, because Waymo has reportedly said riders generally cannot obtain vehicle footage themselves.
Frequently Asked Questions
Who is the man trapped in a self-driving car in the headline?
The headline about a high-tech search for home is tied to coverage by The Anniston Star, 28/22 News and The Jerusalem Post, but we could not verify the rider’s identity or the details. Two other named riders have been reported in separate incidents: Mike Johns in a Waymo and David Moss in a Tesla Cybercab.
How long was the Tesla Cybercab rider trapped?
David Moss was allegedly trapped for 14 minutes on October 5, with about 7 minutes reportedly needed to restart FSD. These figures are allegations and Tesla has not confirmed them in the sources we reviewed.
What did Waymo say about the Phoenix airport incident?
Waymo allegedly said its vehicles have a “pull over” button available to riders and that a software update addressed the issue. We could not verify the statement on the record.
Are robotaxi riders able to stop the car themselves?
Waymo says it offers a pull-over button. For the Cybercab, the sources we reviewed describe remote support stopping the vehicle rather than a rider control, and the car has no steering wheel.
Is NHTSA investigating these specific rides?
Our sources do not say so. NHTSA activity involving Cybercabs is covered in our earlier reports, but nothing we reviewed ties it to the Moss ride.
Can riders get footage after an incident?
Yahoo News reported that Waymo has said passengers generally cannot independently obtain interior or exterior vehicle footage, and that law enforcement may need a search warrant.
Should I avoid robotaxis because of these stories?
That is a personal call. The reports describe delays and stress, not injuries, in the Johns and Moss cases. Knowing the stop control and sharing your trip reduces the risk of a long wait.
