It has been a busy month for autonomous vehicles (AVs). In the first week of September, the policy guidance on AV safety. Although the House bill understandably seeks to promote innovation in the development of AVs, the bill would upset the balance of regulatory authority between the federal government and the states, leaving serious gaps in legal protection of public safety in the near term.

The highly automated vehicles (HAVs)—automated vehicles that could fully control all aspects of driving without human intervention—argue that taking humans out of the driver’s seat could therefore improve vehicle safety by preventing accidents.

On the flip side, many members of the public are not entirely ready to cede control to robot cars. A 2016 study by the revealed that a mere 16 percent of respondents would prefer to ride in a driverless car, while 46 percent would prefer to retain “full control” of a vehicle, and 39 percent would prefer a “partially self-driving” vehicle.

HAVs can introduce new kinds of accidents, including automated system failure; and AVs that require some human monitoring can lead to driver complacency. For example, the concluded that the design of Tesla’s Autopilot system was in part to blame, both because the “automated vehicle control system was not designed to, and could not, identify the truck crossing the Tesla’s path or recognize the impending crash,” and because “the way in which the Tesla ‘Autopilot’ system monitored and responded to the driver’s interaction with the steering wheel was not an effective method of ensuring driver engagement.”

Regardless of whether one favors or fears innovation in autonomous vehicles, there is no question that HAVs—with or without steering wheels—are coming in some form and that they will soon start to share the roads with human drivers, at least for a period of time. When an innovation in business or technology does not fit neatly into a preexisting legislative or regulatory scheme, the innovation can create a policy disruption.

As I have upset the balance of regulatory authority between the federal government and the states—a federalism disruption. In this case, the federalism question and the safety question are intimately related.

Since 1966, the preempts the states from adopting conflicting legislation or regulations. In this scheme, the states maintain regulatory primacy over the “driver,” whom they regulate through insurance rules, licensing, and common law tort rules, as federal law carves out an exception for state tort law. The states can likewise act if NHTSA has not yet adopted a safety standard. HAVs disrupt this federalism balance because vehicle hardware and software can now be programmed in advance to perform tasks formerly within the scope of the driver’s control. HAVs thus blur the distinction between “drivers” and “vehicles,” a key distinction underlying the allocation of regulatory responsibility between states and the federal government for the past 50 years.

To date, despite having the authority to do so, NHTSA has not adopted any mandatory federal safety standards for HAVs. Instead NHTSA has issued a series of informal policy guidance documents that acted as laboratories of policy experimentation, adopting different rules about whether vehicles without a human driver can be tested or deployed on public streets.

Concerned about a possible patchwork of state laws that could hinder innovation, the House has amend the current Motor Vehicle Safety Act to address HAVs explicitly.

Although H.R. 3388 appears to do a good job of promoting innovation, it fares less well on the connection between federalism and safety. In a departure from the current balance of power between the federal government and the states, the bill would immediately preempt any state or local government from adopting or enforcing “any law or regulation regarding the design, construction, or performance of highly automated vehicles, automated driving systems, or components of automated driving systems,” even in the absence of a federal motor vehicle safety standard.

This immediate preemption may create a safety gap that could persist for some time. The bill would require auto manufacturers to submit a safety assessment letter as contemplated by any NHTSA guidance in effect that requires such an assessment.

However, the current 2017 NHTSA guidance does not require such a certification letter at all. Instead, the guidance prohibit NHTSA from conditioning the “deployment or testing” of HAVs “on review of safety assessment certifications.”

H.R. 3388 appears to leave a safety gap until NHTSA adopts a rule on safety assessments. It would further adopt specific, binding federal safety standards. This, too, leaves a potential safety gap that the states could not fill with legislation or regulations. Preemption of state safety rules in the absence of federal motor vehicle safety standards represents a significant change from the status quo.

The Self-Drive Act’s preemption language expressly carves out both state common law and state or local laws on “registration, licensing, driving education and training, insurance, law enforcement, crash investigations, safety and emissions inspections, congestion management of vehicles on the street,” as long as those laws do not pose “an unreasonable restriction on the design, construction, or performance of highly automated vehicles, automated driving systems, or components of automated driving systems.”

Although state tort law may create incentives for manufacturers to address safety concerns once courts have ruled on the appropriate standard of care, lawsuits can be slow, and they are retrospective in nature. Courts can only respond to accidents after they occur. Although avoiding a patchwork of state laws is a reasonable way to promote innovation and economies of scale, preempting state action even in the absence of federal safety standards fails to take into account the significant concerns that states have in protecting their citizens from harm as a result of the deployment of HAVs on public roads.

Despite NHTSA’s “strong encouragement” that the states not acknowledges that states can continue to regulate until the federal government adopts binding requirements. However, if the 2017 voluntary guidance were to be combined with the premature preemption of state law proposed by H.R. 3388, this potent combination of voluntary standards and premature preemption would leave unnecessary gaps in HAV safety regulation.

As the Senate takes up the issue, that chamber should adopt a different approach that better balances the promotion of innovation and with the consideration of safety concerns. A better approach would permit the states to continue to engage in policy experimentation on HAV safety regulation until the federal government adopts uniform federal motor vehicle safety standards – consistent with the current regime under the Motor Vehicle Safety Act.

At the very least, in the alternative, Congress could create a two-car system, at least temporarily until NHTSA adopts uniform federal motor vehicle safety standards. The two-car system could be exempts the state of California from this limitation if the state demonstrates that its rules are at least as protective of human health as the federal standard. Other states can then adopt California’s emissions standard, creating a two-car regime.

Maintaining jurisdictional overlap in this dynamic fashion led to policy improvement over time in the emissions context. Such a waiver provision and two-car system could be applied to HAV safety regulation and yield similar results, at least on a temporary basis. We already live in a two-car regime for emissions control—extending that regime to HAVs would offer a sensible middle ground.