Why the Safety of Self-Driving Cars Matters Right Now
Self-driving cars are still a novelty on most roads worldwide, but in San Francisco, Phoenix, and Los Angeles they have already become part of everyday traffic. Waymo, the robotaxi subsidiary of Google parent Alphabet, now carries hundreds of thousands of riders a week with nobody sitting behind the wheel. For anyone tracking electric vehicle technology, that is more than an American curiosity. Driver-assistance systems in current EVs keep getting more capable year after year, and the question of how safely a machine actually makes decisions in traffic increasingly touches the entire spectrum of SAE levels of autonomous driving, from basic assistance all the way to genuine self-driving capability.
The core difficulty in judging this technology is that "safety" carries two very different meanings here. On one side sit cold statistics about crash rates. On the other sit individual, often dramatic incidents that light up news feeds and social media. Both are relevant, and both get used as ammunition against the other side of the argument. A recent report from CNN tackled exactly that tension: reporter Clare Duffy climbed into a driverless Waymo for her podcast "Terms of Service" and simultaneously pressed a Waymo executive for the numbers behind the system. The facts discussed in that video offer a solid foundation for a nuanced look at the safety question around Waymo and comparable systems, instead of reducing the debate to a single viral clip.
Source: CNN – Are self-driving cars safe? We got into a Waymo to find out | Terms of Service
The Waymo Numbers in Detail: What the CNN Report Shows
At the heart of the video is a specific figure that Waymo itself publishes: according to the company's own data, human drivers cause injury-causing crashes at roughly 13 times the rate of Waymo's autonomous vehicles, and that figure explicitly includes collisions involving pedestrians. For a company that lives under constant public scrutiny, that is a deliberately prominent number to lead with. Even so, it comes from an operation that has now logged an enormous number of miles, and US regulators do review at least part of the underlying data.
A 13x Lower Rate of Serious Crashes: What the Statistic Actually Means
What matters here is how a number like this is built. It does not compare "crash versus crash" in a raw sense, it compares the frequency of collisions that actually result in injury, measured against distance driven. Minor fender-benders or incidents with no injury are weighted differently than serious collisions. That distinction is exactly what makes the 13x gap meaningful: it describes real-world outcomes on public roads, not a theoretical error rate buried in the software. The technical backbone behind those outcomes is a combination of lidar, radar, and camera sensors that constantly gather data and process it in real time, catching dangerous situations earlier than a human eye typically could.
The Incidents That Fuel Skepticism
At the same time, the CNN report does not gloss over the downsides. Multiple documented cases show Waymo vehicles blocking emergency access routes because the system misread an unusual traffic situation. Other incidents involved vehicles illegally passing stopped school buses, even though a strict no-passing rule applies to every driver in that situation. The case that drew the most attention in San Francisco was the death of "KitKat," a beloved neighborhood bodega cat struck and killed by a Waymo vehicle. Incidents like these are statistically rare but emotionally significant, and they expose a fundamental weakness of automated systems: rare, thinly trained edge cases remain a vulnerability no matter how strong the average safety record looks.
- 13x lower rate of injury-causing crashes according to Waymo's own data
- Figure explicitly includes collisions with pedestrians
- Documented incidents: blocked emergency access routes
- Documented incidents: illegally passing stopped school buses
- High-profile case: death of bodega cat "KitKat" in San Francisco
Cabin Monitoring: What the Cameras Record During the Ride
Another segment of the video looks at how Waymo monitors rider behavior inside the vehicle. According to the Waymo executive interviewed, interior cameras record passengers throughout the ride, partly for rider safety, partly to resolve disputes, and partly to improve the service overall. For riders, that is a trade-off: more accountability and traceability if something goes wrong, but noticeably less privacy than in a personal car. In Germany, where data-protection rules are traditionally applied more strictly than in much of the US, this is likely to become one of the first sticking points regulators and operators would need to resolve before any large-scale robotaxi rollout could happen there.
Expansion as the Real Stress Test for Safety Claims
The video also touches on Waymo's plans to expand into more US cities and markets. Every new city brings different road layouts, weather patterns, and driving cultures that the system initially knows little about. Scaling up is therefore the real test of any safety claim: a fleet that performs well in one familiar city still has to prove it can hold the same accident record somewhere entirely new. This is exactly where autonomous fleets increasingly diverge from traditional taxi and ride-hailing services, since human drivers adapt intuitively to unfamiliar environments while software first needs fresh training data before it can do the same.
Outlook: What the Numbers Mean Beyond the US, and for the Future
The data discussed in the video comes from the United States, a market with its own legal framework, its own driving culture, and a baseline crash rate that runs considerably higher than in much of Europe. That does not diminish the 13x gap, but it does make a direct one-to-one comparison to other countries difficult. Anyone trying to make sense of these numbers should also understand the technical foundation behind them, including how modern systems fuse sensor data from multiple sources to reach a driving decision in the first place, and how much that approach still varies from one manufacturer to the next.
Regulation and Approval: The German Approach
In Germany, the legal framework for highly automated and fully autonomous driving is already written into law, but in practice it is applied far more cautiously than in parts of the United States. Approval, testing, and liability questions run through a tight collaboration between federal authorities, the KBA (Federal Motor Transport Authority), TÜV inspection bodies, and a handful of designated pilot regions. Before a robotaxi system like the one shown in the video could operate at scale there, it would need to clear safety requirements that go well beyond a simple crash-rate comparison. Readers who want the legal detail on who actually pays after a crash can find it in our companion piece on liability after a self-driving car accident.
What Happens to Taxi and Ride-Hailing Jobs
Another topic the video raises directly is the future of traditional ride services. The more cities that approve robotaxi fleets, the more pressure builds on the business model of taxi companies and ride-hailing drivers. In the US, that shift is already visible in cities where Waymo runs at scale: in San Francisco, for example, Waymo has overtaken Lyft in gross bookings within its own service area, while its former robotaxi partnership with Uber in Phoenix has wound down as Waymo shifts riders onto its own app. Elsewhere, that effect is still a long way off, but the debate over jobs in passenger transport is likely to gain momentum in step with the technology, especially once early pilot regions start offering similar services of their own.
Autonomous Vehicles and Charging Infrastructure Go Hand in Hand
For a site built around charging infrastructure, one detail stands out: nearly every robotaxi fleet operating today runs entirely on electric power, partly because electric drivetrains are simply easier to pair with the heavy computing loads that sensors and self-driving software demand. A growing fleet of autonomous vehicles therefore means a growing, steady demand for fast, reliable charging, ideally paired with automated docking at the charger with no human involvement at all. Anyone who wants to brush up on the basics of EV charging independently of all this can start with our broader guide to fast-charging an EV.
- US crash statistics don't translate one-to-one to other countries
- In Germany, approval runs through TÜV, the KBA, and defined pilot regions
- Liability rules already exist on paper but remain largely untested in court
- The taxi and ride-hailing industry faces a long-term structural shift
- Robotaxi fleets are almost entirely electric, with knock-on effects for charging demand
In the end, the picture is a sober one. The numbers Waymo presents in the CNN report make a strong statistical case for autonomous systems, while individual, well-documented incidents show that residual risk remains, especially in rare edge cases the software has barely seen before. For anyone following EV technology, that argues for watching two tracks at once: keep an eye on how sensors and software keep improving, and pay close attention to the political and legal debate in your own country before robotaxis become part of everyday life there too.
Additional Video
For a deeper look beyond the raw safety numbers, this video from tech creator Shelby Church digs into the overlooked risks, liability gaps, and edge cases that rarely make it into the headline statistics on self-driving cars.
Source: Shelby Church – The Problem With Self Driving Cars Nobody Is Talking About
Frequently Asked Questions
Are self-driving cars actually safer than human drivers?
According to Waymo's own data, human drivers cause injury-causing crashes at roughly 13 times the rate of Waymo's autonomous vehicles, including crashes involving pedestrians. Independent studies point to a similar trend, though with a much smaller data set than traditional crash statistics provide.
What happened in the high-profile incidents involving Waymo vehicles?
Documented cases include blocked emergency access routes, vehicles illegally passing stopped school buses, and the death of a cat in San Francisco. These cases are rare, but they show that unusual edge cases remain a genuine weak spot for autonomous systems.
Are self-driving cars allowed to operate in Germany?
Yes, but only on a limited basis, within defined pilot regions and under supervision from authorities like the TÜV and the KBA. There is currently no widespread robotaxi service in Germany comparable to what runs in San Francisco.
Who is liable when a self-driving car causes an accident?
Liability depends on the level of automation involved and the specific circumstances of the crash, and it shifts increasingly from the driver toward the manufacturer or operator as autonomy increases. We cover the legal details in full in our dedicated article on liability.
