Waymo

1. Waymo

Waymo is the autonomous driving company owned by Alphabet Inc., the parent company of Google. It was originally founded in 2009 as the Google Self-Driving Car Project and has since become one of the leading developers of self-driving cars. Customers can request a ride and travel to their destination without a human driver. The service currently operates in several cities in the United States. As of now, they are testing their vehicles on London roads with trained human safety drivers for a possible expansion in late 2026.

Website: waymo.com.

2. The AI technologies Employed

Waymo develops autonomous driving technology through two main systems: the Waymo Driver and Driver Assist Technology ^[1]. The Waymo Driver is the company's fully autonomous driving systems that allows passengers to travel without driving the vehicle themselves. The system uses artificial intelligence (AI), detailed digital maps, 29 cameras on the vehicle, a radar, and LiDAR sensors to understand its surroundings. LiDAR (Light Detection and Ranging) helps create a 3D picture of the environment around the vehicle. Sensors placed around the car send out millions of laser pulses and measure how long they take to bounce back from nearby objects. This allows the vehicle to detect pedestrians, cyclists, and other vehicles. The AI system combines information from these sensors with detailed maps to determine the vehicle's location and understand what is happening around it. It then decides how the vehicle should respond by choosing routes, adjusting speed, changing lanes, braking, or steering. The system has been trained using millions of miles driven on public roads and billions of miles in computer simulations.

Waymo also offers Driver Assist Technology. Unlike the fully autonomous Waymo Driver, this system still requires a human driver to pay attention to the road and be ready to take control when necessary. The technology assists with driving tasks but does not replace the driver's responsibility for operating the vehicle.

3. Ethical concerns

Waymo shows substantial awareness of the ethical and safety challenges tied to autonomous vehicles. The company promotes its extensive training and testing in simulations and real roads. They have added many additional safety features, such as backup steering and braking systems, extra collision-avoidance systems, secondary computers, independent power supplies, and strong cybersecurity protections ^[1:1]. These efforts show a strong focus on safety and reducing risk. However, since the system still relies on AI to make driving decisions, there are ongoing ethical concerns about accountability, safety, and transparency of this technology. It is not always clear who is responsible when something goes wrong, or how exactly the system makes decisions in rare or dangerous situations.

3.1 Safety through third parties

Waymo places a strong emphasis on safety, this is reflected in its extensive testing and training processes. However, real-world investigations and reported incidents still raise concerns. For example, CNN and other reports have described cases where Waymo vehicles ran red lights, entered flooded areas, blocked emergency responders, or were involved in a software recall after entering closed highway construction zones ^{[2] [3]}. Even if these incidents are rare compared to the total number of miles driven, they show that autonomous systems can still fail in complex and unpredictable situations, sometimes with serious consequences.

Waymo reports strong safety performance compared to human drivers, but these figures are largely based on company-selected data, which limits how independently they can be assessed. Waymo's own published safety data claims 92% fewer serious injury crashes and 83% fewer airbag deployments in crashes versus human-driving benchmarks in the same cities ^[4], compared against the rawer NHTSA picture: 1,429 reported crashes between mid-2021 and late 2025, with 117 injuries and 2 fatalities (though most of these involved other drivers hitting Waymo vehicles rather than Waymo causing the collision) ^[5]. In the United States, oversight has been inconsistent and mostly happens after problems occur. NHTSA requires companies to report most crashes, but its safety reviews are voluntary, meaning the company writes its own report and NHTSA doesn't have to approve or check it ^[6]. In California, they have started fixing this: since April 2026, new DMV rules require companies to submit a detailed "safety case" that the DMV actually checks (and can have outside experts check too) before giving a permit, instead of just trusting the company's own reporting ^[7].

3.2 Not transparent enough

Waymo shows a relatively high level of transparency in research compared to many companies in the autonomous vehicle industry. For example, it publishes the Waymo Open Dataset on GitHub and has released dozens of research papers on perception, prediction, and safety in autonomous driving ^{[8] [9]}. This helps researchers and the wider public better understand the technical foundations of self-driving systems. However, this form of transparency is mainly focused on controlled datasets and academic research rather than real-world operational decision-making. The Open Dataset and research papers are also a recruiting and credibility tool for Waymo, which doesn't make the openness fake, but it's worth flagging that "transparency" here serves the company's interests too, not just the public's. Key aspects such as exact safety thresholds, failure cases in complex environments, and how trade-offs are made in unusual cases are still not fully disclosed. As a result, there is a gap between technical openness and practical transparency about how the system behaves in real driving situations, especially when safety-critical decisions are made.

3.3 Who's accountable when something goes wrong?

These issues raise broader questions about responsibility. When a self-driving car makes a mistake, it is not always clear who is accountable, the software, the engineers who built it, or the company that deployed it. This creates an ethical challenge because, while autonomous systems are designed to reduce human error, they also spread decision-making across complex systems that are harder to fully understand and regulate. Allen’s idea of accountability helps explain this problem ^[10]. As AI takes over more driving decisions, it becomes harder to decide who should explain and take responsibility for harm. Her view is that technology should not remove responsibility, but instead accountability systems should adapt so that responsibility is still clear even when AI is involved. We need new rules, structures, or frameworks that clearly assign responsibility even in situations where an AI made the proximate decision.

In the case of a crash involving Waymo, responsibility would still ultimately lie with the company and the developers who created and deployed the system, even if the decision was made by the AI. Accountability is not only about preventing accidents, but also about explaining what happened when something goes wrong. People affected by an incident may expect clear explanations of how the system saw the situation, assessed risk, and made its decision. Waymo hasn't committed to a standard for post-incident explanation the way it has committed to a standard for pre-deployment testing.

4. Recommendations

To address the concerns highlighted above, Waymo could strengthen its approach to safety, transparency, and accountability in several ways.

1. Improving safety through stronger external standards: Waymo currently works under the US system, which allows faster testing and deployment but relies a lot on company reports and checking problems after they happen. This means there is less independent approval before the cars are used on the road. Waymo could follow stricter international rules instead of relying on the mostly voluntary US system. The UN's UNECE group has written a new global rule, expected to be approved in June 2026, that requires companies to prove their self-driving system is safe through detailed evidence, plus ongoing monitoring after the cars are on the road ^[11]. The EU's AI Act also sets extra rules for "high-risk" AI systems like this one. If Waymo followed standards like these, its safety claims would be checked by someone other than itself.

2. Increasing practical transparency: Waymo could go beyond publishing research papers and datasets by sharing more real-world safety information. This could include clearer reports about serious incidents, more explanation of how the system handles difficult situations, and regular reports on performance in complex driving conditions. Full access to all internal systems is not realistic, but more structured reporting and independent review would help people better understand how safe the system really is in practice.

3. Making accountability clearer: Waymo should also clearly explain who is responsible when an accident happens during self-driving mode. In most cases, responsibility lies with the company rather than the passenger, but this is not always clearly stated in Waymo’s public information.