Are there any 100% self-driving cars?

No—there are currently no cars available to the public that can drive themselves everywhere, in all conditions, without human oversight. What does exist are limited, geo-fenced robotaxi services that operate without a human driver in specific areas and conditions, plus consumer vehicles with advanced driver-assistance that still require active human supervision or only work in narrow scenarios.

What “100% self-driving” really means

When people say “100% self-driving,” they typically mean SAE Level 5 autonomy: a vehicle that can handle all driving tasks, in any environment and weather, with no human intervention or operational restrictions. No company has achieved Level 5, and none sells such a vehicle. Most real-world deployments today fit SAE Level 4 (driverless only within defined operational design domains) or lower (Level 2–3, which still rely on the human). Understanding this distinction helps separate marketing claims from the technical and regulatory reality.

Where fully driverless rides do exist—and their limits

A handful of companies operate driverless ride-hailing in constrained areas and conditions. The list below highlights prominent deployments to illustrate what’s real today and what’s still experimental.

• Waymo (U.S.): Offers driverless robotaxi service without a human driver in the vehicle within mapped parts of the Phoenix metro area, and has expanded service areas in other U.S. cities over time, subject to state and local approvals. Operations and service footprints can change as regulators review safety data.
• Baidu Apollo Go (China): Provides driverless rides in designated zones of cities such as Wuhan and parts of Beijing (e.g., Yizhuang), with service areas, times, and road types defined by local permits.
• AutoX (China): Has run driverless operations in parts of Shenzhen, again within geo-fenced areas and under specified conditions.
• Zoox (U.S.): Tests purpose-built, steering-wheel-free shuttles without a driver on limited routes; as of now, public access remains restricted and operations are tightly controlled.
• Cruise (U.S.): Paused driverless service after a 2023 incident and has been working through regulatory reviews and staged testing. Any return to public driverless service remains limited and contingent on regulators and company readiness.

These services demonstrate Level 4 capabilities: no driver in the front seat, but only inside carefully mapped, geo-fenced zones and often with restrictions on speed, time of day, weather, and road types.

What you can actually buy: today’s consumer systems

Consumer vehicles on sale today do not drive themselves everywhere. The systems below show the highest levels available to drivers—and their constraints.

• Level 2 (assistance with supervision): Widely available systems such as Tesla FSD (Supervised), GM Super Cruise/Ultra Cruise, Ford BlueCruise, and others automate steering, acceleration, and braking but legally and technically keep the human responsible at all times.
• Level 3 (conditional automation in narrow scenarios): Mercedes-Benz Drive Pilot is approved in limited regions of the U.S. (e.g., California and Nevada) for low-speed, daytime highway traffic under specific conditions; the system can handle the driving, but only within those boundaries. Other Level 3 offerings exist in select countries (e.g., Honda in Japan, BMW in parts of Europe) with similar narrow use cases.

In short, you can buy driver-assistance and, in some markets, narrowly constrained Level 3 functionality—but not a car that replaces the driver universally.

Why Level 5 remains out of reach

Moving from impressive demos to a car that can truly drive anywhere is a leap that spans technology, safety, and public policy. These factors explain the gap.

• Edge cases and unpredictability: Rare, complex scenarios (unprotected turns, unusual construction zones, erratic human behavior, adverse weather) are hard to handle with the reliability demanded for driverless operation everywhere.
• Sensing and redundancy: Achieving human-level perception across all conditions requires robust sensor suites, fail-operational computing, and redundant actuation—raising complexity, cost, and engineering hurdles.
• High-definition maps and generalization: Many Level 4 systems lean on detailed maps; keeping them current at national scale is nontrivial. Purely “map-light” approaches still struggle with full generalization.
• Safety validation: Proving statistically significant safety across the long tail of events demands enormous real-world exposure, simulation, and transparent reporting.
• Regulatory and liability frameworks: Rules differ by city, state, and country, shaping where and how driverless operations can scale.
• Unit economics: Robotaxis must balance hardware costs, maintenance, remote assistance, and fleet operations to be viable at scale.

These challenges are being chipped away by iterative software improvements, better sensors and compute, and expanded testing—but none has been solved comprehensively enough to unlock Level 5.

How to assess claims you’ll hear

Marketing can blur lines. Use the checklist below to evaluate whether a system is truly driverless—and where it can operate.

• Supervision required? If the driver must pay attention at all times, it’s not self-driving (that’s Level 2).
• Operational design domain: Ask where it works (which roads), when (time of day, weather), and at what speeds. The narrower the domain, the less it resembles “anywhere, anytime.”
• Regulatory status: Paid driverless service typically requires explicit permits; check local transportation or motor vehicle regulators.
• Public access vs. pilots: Employee-only or invitation-only programs are steps toward scale, not broad availability.
• Independent safety data: Look for crash reports, disengagement statistics where available, and documented recalls or software updates.

Applying these filters quickly clarifies whether a claim describes a supervised driver-assist feature, a geo-fenced robotaxi, or the still-hypothetical Level 5.

The bottom line for 2025

There are no consumer-available Level 5 vehicles and no driverless cars that can go anywhere, anytime. Limited driverless robotaxi services (Level 4) operate in specific cities and zones, and consumer cars top out at Level 2 assistance or tightly constrained Level 3. Progress continues, but “100% self-driving” remains a milestone that the industry has not yet reached.

Summary

No, there are no 100% self-driving (SAE Level 5) cars available today. A few companies offer driverless robotaxis within geo-fenced areas under strict conditions, while all consumer vehicles either require constant driver supervision (Level 2) or provide limited conditional automation (Level 3) in narrow scenarios. Technical edge cases, safety validation, regulation, and economics are the main hurdles between today’s deployments and true “drive anywhere” autonomy.

Can self-driving cars ever be 100% safe?

No. Self-driving cars will never be 100% trustworthy. Once they are 99.8% trustworthy, they will become commonplace. Once they reach 99.97% trustworthy, they will be the norm. In contrast, human-driven cars will remain around 99.7% trustworthy, and will likely never exceed that level.

How much does a fully self-driving car cost?

Self-driving car costs vary significantly, but for current advanced systems, prices can add $8,000 to over $10,000 to a car’s price, while higher-end systems like those from Waymo can bring the total cost to $200,000 or more. However, costs are expected to decrease dramatically over the next decade due to economies of scale, with some projections suggesting that fully autonomous vehicles could eventually cost less than $30,000. Fleet ownership and operating costs are also expected to become cheaper than traditional vehicles. 
Current Costs (2025)

• Additional Hardware: Features for supervised self-driving, such as Tesla’s Full Self-Driving (FSD) package, can cost around $8,000 to $10,000 as an add-on to a base vehicle price of about $40,000+. 
• High-End Systems: High-end, fully autonomous systems, like those used in Waymo’s fleet, involve a significant investment in sensors and computers, potentially making the total vehicle cost $200,000 or more. 
• Subscription Models: For existing vehicles, a subscription model for self-driving capabilities, like Mercedes-Benz’s Drive Pilot, can cost around $2,500 in addition to the vehicle’s purchase price. 

Future Cost Reductions

• Reduced Component Costs: Opens in new tabThe cost of specialized hardware, such as lidar, has already decreased significantly due to advancements and scaling. 
• Tesla Robotaxi: Opens in new tabTesla CEO Elon Musk has stated their future Robotaxi (or Cybercab) will cost less than $30,000 when production begins in 2026. 
• Fleet Operating Costs: Opens in new tabAutonomous vehicle fleet ownership and operating costs are expected to be cut in half by 2030-2040 compared to traditional vehicles. 

Key Factors Influencing Future Costs

• Economies of Scale: Widespread adoption will lead to increased production and lower manufacturing costs for sensors, cameras, and computers. 
• Technological Advancements: Continued research and development will improve sensor technology and reduce the reliance on expensive components. 
• Market Competition: As more companies enter the autonomous vehicle market, competition will drive down prices for both the vehicles and the technology they use. 

What states allow full self-driving cars?

Twenty-nine states—Alabama, Arkansas, California, Colorado, Connecticut, Florida, Georgia, Illinois, Indiana, Kentucky, Louisiana, Maine, Michigan, Mississippi, Nebraska, New York, Nevada, North Carolina, North Dakota, Oregon, Pennsylvania, South Carolina, Tennessee, Texas, Utah, Virginia, Vermont, Washington and …

Are there any fully self-driving cars?

As of late 2024, no system has achieved full autonomy (SAE Level 5). In December 2020, Waymo was the first to offer rides in self-driving taxis to the public in limited geographic areas (SAE Level 4), and as of April 2024 offers services in Arizona (Phoenix) and California (San Francisco and Los Angeles).