AV Regulation in 2025: Where the Laws Are, Where They're Missing, and What's Blocking Deployment

Waymo operates commercially in San Francisco, Phoenix, and Los Angeles. Tesla claims Autopilot and Full Self-Driving are semi-autonomous features under driver supervision. Aurora launched commercial autonomous trucking on Texas highways in April 2024. All of this is happening under regulatory frameworks that weren't designed for it, and in most cases still aren't.

The gap between what's technically deployed and what's legally governed is one of the more interesting fault lines in the current automotive landscape.

## Where the Laws Are

The United States has no federal autonomous vehicle law. Congressional attempts — the AV START Act, various subsequent drafts — have repeatedly stalled. What exists instead is a patchwork of state-level frameworks, NHTSA voluntary guidance, and exemption processes.

California is the most active regulator. The DMV issues both testing permits and commercial deployment permits. Waymo's robotaxi service operates under a commercial deployment permit that required years of data submission and review. Cruise — GM's autonomous vehicle unit — had its permit revoked in 2023 after a pedestrian accident and a subsequent disclosure failure; the regulatory response demonstrated both that permits can be pulled and that the disclosure requirements are serious.

Texas and Arizona operate with minimal state-level restrictions. Both Waymo and Uber's autonomous vehicle programs have used these states for operational development precisely because the regulatory environment is permissive. Aurora chose Texas I-45 for its first commercial autonomous trucking launch partly for this reason.

In Europe, Germany's Autonomous Driving Act (2021) and subsequent regulations allow Level 4 autonomous operation in defined areas. China has moved aggressively, with Baidu's Apollo Go robotaxi service operating in multiple cities. The EU is working on updated type approval regulations, but cross-border autonomous vehicle frameworks don't yet exist.

---

## Where the Laws Are Missing

The liability gap is the most consequential blank space. When an autonomous vehicle causes an injury, the existing frameworks — designed around driver negligence — don't map cleanly to software-controlled vehicles. Current law in most jurisdictions defaults to the vehicle owner as liable, which creates an awkward situation where Waymo riders are technically liable for accidents in a vehicle they're not controlling.

California's AB 2286 (2024) attempted to establish clear liability for commercial robotaxi operators, placing primary liability on the permit holder rather than the passenger. Similar legislation is pending in other states. But the federal gap means liability frameworks vary dramatically by geography.

*Insurance* is the related unsolved problem. Commercial AV operators currently self-insure in most cases — Waymo's agreement with the insurance market is reportedly internal, and the premiums required for third-party coverage of a new risk category without actuarial history are prohibitive. As deployment scales, this will force either sector-specific insurance products or some form of industry pooling.

Data recording and disclosure requirements are inconsistently defined. Most jurisdictions require AV operators to report crashes. Fewer require disclosure of disengagements (moments where autonomous systems cede control to a human). Fewer still require disclosure of system performance data that would allow independent assessment of safety improvements over time.

---

## What's Actually Blocking Deployment

The technology is not blocking deployment — at least not for the limited operational design domains that Waymo, Cruise, and Aurora are targeting. The blocking factors are:

**Regulatory latency**: permitting processes designed for the validation of small fleets don't scale efficiently to tens of thousands of vehicles. Waymo's expansion from San Francisco to Los Angeles took years of separate regulatory engagement.

**Public acceptance**: Cruise's 2023 accident and the public and regulatory response demonstrated that a single high-profile incident can freeze an industry's expansion. The political economy of autonomous vehicles is fragile.

**Unit economics**: Waymo's vehicles cost substantially more to manufacture than a conventional vehicle, and the cost of remotely monitored autonomy infrastructure doesn't yet reach positive margin at current scale. The path to profitability requires volume that the current regulatory environment constrains.

---

## The Verdict

Autonomous vehicle deployment is happening, but in a form that looks nothing like the "2020 fully autonomous vehicles everywhere" predictions of 2015. Waymo's operational profile — geofenced, weather-constrained, expensive, commercially limited — is impressive engineering and commercially marginal. The regulation isn't the only reason, but it's a significant constraint. Federal clarity on liability, mandatory safety data disclosure, and insurance frameworks designed for AV risk would accelerate the deployment curve. The numbers don't lie: without those, the market stays small and regional for the medium term.

AV Regulation in 2025: Where the Laws Are, Where They're Missing, and What's Blocking Deployment

// COMMENTS

ON THIS PAGE