Autonomous Driving Liability Law 2026: Who's Responsible When the Car Decides?

On April 8, 2024, a **Waymo** robotaxi in San Francisco struck a bicyclist who ran a red light. The vehicle had no safety driver. **Waymo** was the operator. The bicyclist's injuries were minor, but the legal question was significant: when a vehicle with no human driver causes harm, who is liable?

The insurance claim was resolved privately. The legal precedent was not set. That ambiguity — which persists through 2026 across every major automotive market — is more consequential for autonomous vehicle deployment timelines than any remaining sensor or software challenge.

## The Numbers

| SAE Level | Description | Liability | Examples |
|-----------|-------------|-----------|---------|
| L2 | Driver must monitor, system assists | Driver responsible | **Tesla** Autopilot, **GM** Super Cruise |
| L3 | System handles driving, driver must be available | Disputed — shifting toward OEM | **Mercedes** Drive Pilot |
| L4 | Fully autonomous in defined conditions | OEM/operator | **Waymo** geo-fenced zones |
| L5 | Full autonomy everywhere | OEM | Not commercially deployed |

The liability boundary between L2 and L3 is the most contested. At L2, the human driver is legally responsible — even if the automation fails, the driver was supposed to be monitoring and intervening. At L3, the system is handling the driving task and the human is not required to monitor continuously. When an L3 system fails, the human may not have had enough time to take over — and the question of who bears liability for that failure is unresolved in most US jurisdictions.

## Why Tesla Avoids Level 3

**Tesla**'s Full Self-Driving package remains SAE Level 2 by the company's own classification and regulatory treatment. The driver is required to maintain attention and hands on the wheel. This is not primarily a technology decision.

**Tesla** has explicitly stated in regulatory filings that its current system does not meet the standard for a driver to be safely unattended. More importantly, L3 certification would require **Tesla** to accept liability for failures that occur during the autonomous driving mode — the same liability that **Mercedes-Benz** accepted when it received L3 certification for its Drive Pilot system in Nevada and California.

**Mercedes-Benz** became the first OEM to receive L3 certification in the US in 2023. The certification documents explicitly state that **Mercedes** accepts legal responsibility for incidents occurring while Drive Pilot is engaged at speeds below 40 mph on approved highway segments. **Mercedes** priced Drive Pilot as a premium subscription in part to offset the liability exposure.

**Tesla** avoids this by keeping FSD legally at L2. The customer bears responsibility. When crashes occur during FSD engagement, **Tesla**'s legal position has consistently been that the driver failed to maintain attention.

## The Waymo Liability Position

**Waymo**'s L4 commercial deployment has the clearest liability structure: **Waymo** is the operator, there is no human driver, and **Waymo** is unambiguously responsible when something goes wrong. This is simpler in a legal sense — one party, one liability — even though the incidents that occur are more novel.

**Waymo** has worked with Lloyd's of London on specialized AV insurance policies that cover its commercial deployments. The coverage structure acknowledges the absence of a human driver and prices the risk accordingly.

**GM**'s Cruise subsidiary provides the most instructive cautionary case. Following a 2023 incident in San Francisco where a Cruise robotaxi struck a pedestrian and the vehicle pulled over on top of the pedestrian rather than calling for help, GM suspended Cruise operations and eventually shut down the robotaxi division. The incident's direct cause was a software failure in post-incident behavior — not the initial strike — but the regulatory response was comprehensive.

## The Verdict

The liability uncertainty is, in practice, a commercial deployment bottleneck. Automakers willing to accept L3 liability are constrained to scenarios where they can precisely define the conditions of operation. Fully autonomous L4 operators have cleaner liability structures but face higher ongoing insurance costs and regulatory scrutiny. L2 manufacturers avoid liability but face increasing NHTSA attention to the gap between marketing language ("Full Self-Driving") and legal responsibility.

State-level legislation in California, Nevada, and Texas defining operator liability frameworks, and the NHTSA's forthcoming AV regulatory framework will collectively move the liability question from ambiguous to defined — on a 3–5 year timeline. Until that happens, the commercial deployment window for any system more ambitious than **Waymo**'s carefully geo-fenced operations will remain constrained less by technology than by paperwork.

The numbers don't lie: 25 million Waymo autonomous miles. The litigation landscape doesn't lie either: still no settled legal precedent for what happens when the car is wrong.

Autonomous Driving Liability Law 2026: Who's Responsible When the Car Decides?

// COMMENTS

ON THIS PAGE