Arizona State University Guest Blog: Policymakers tread the swirling waters of AI innovation
By Kris Hartley
In May 2025, Republican leaders in the U.S. House of Representatives advanced a legislative proposal to enact a 10-year moratorium on state-level regulation of artificial intelligence (AI). While the proposal failed in the Senate, the effort signals an emerging fault line in views about how AI regulation is balanced between the state and national level – and a turn towards policy nationalization by a Republican party that has spent decades advocating mostly for the opposite.
Rapid advancements in AI have led to questions about appropriate regulatory scope, whether national or local. Private sector innovation often outpaces policy responses, which need time for observation, design, and implementation. AI is a notable case given the complexity of its underlying technology. The immediate and universal impact of AI on daily life also raises the political visibility of policy interventions. Ensuring that the growth of AI is safe, equitable, and sustainable is one of the major policy challenges of the 2020s.
AI is an abstraction to many non-experts but manifests itself in practical ways that are widely relatable. Aside from accessible content-generation applications like ChatGPT, one increasingly common AI application is autonomous vehicles (AVs) – those that substitute automated driving systems for direct human input. Autonomous ‘robotaxis’ have been in development for over two decades and are now common in Phoenix, Los Angeles, San Francisco, Austin, and many cities in China.
More than a dozen companies are currently testing AVs in Arizona, accounting for over 600 vehicles. An early player in the autonomous robotaxi market, Waymo (a subsidiary of Google’s parent company Alphabet) has been at the forefront of AV testing and rollout since its establishment in 2016. The company’s recentered focus from R&D to commercialization and scalability culminated in its 2020 service launch in the Phoenix area.
Waymo’s Phoenix rollout marked a milestone that was not only technological; it was the product of a governance effort that prioritized experimentation, with regulatory restrictions focusing primarily on safety and legal liability. Given that policy issues concerning AVs extend to multiple domains (e.g., data privacy and physical planning), Arizona’s innovation-first approach can be considered an aggressive strategy.
This permissive policy framework was complemented by an effort to avoid regulatory contradictions. According to Brian Norman of Arizona’s Goldwater Institute, “In Arizona, regulatory authority is streamlined at the state level, ensuring safety while avoiding the policy fragmentation that too often creates complicated webs of regulation.” The AV policy strategy that Norman describes as ‘permissionless innovation’ and a ‘freedom-based model for technology’ reflects a light-touch approach that is itself innovative and reactive: act only when problems arise.
What lesson does Arizona’s experience offer for AI and AV policy elsewhere? Two issues define this context. First, AV is a rapidly evolving policy domain with complex technological dimensions not fully understood by the common user or legislator. Second, Arizona’s pro-growth strategy seeks to lure industry investment from regional competitors (i.e., California). A comparison is Covid-19, characterized by (1) a rapidly evolving phenomenon with complex and potentially long-term scientific dimensions that were not fully understood when policy action was undertaken, and (2) the prospect that strong policy intervention could stunt economic activity. In situations where the risk of error is deemed acceptable, pilot cases or demonstration projects can establish an evidence-based case for policy action. However, brief experiments in the face of rapid change struggle to produce insights on medium- or long-term impacts. The adoption of policies for continually shifting issues like technological evolution and public health always carries uncertainty.
The appetite for expeditious regulation can compel legislatures to grant the private sector first-mover status on complicated technical issues. For example, early in Arizona’s AV regulatory journey, a legislator was quoted in the New York Times as saying that “he would wait for the insurance industry to guide regulators on liability policies for driverless cars, amid questions about who is responsible in a crash if the car isn’t driven by a human.” Such regulatory deference is common where expert input is needed, but the consequences of policy failure (e.g., risk to public health) raise the political stakes. Among the many issues that governments must consider in addressing rapidly evolving technical and scientific issues are the capacity for regulatory flexibility and broader societal appetite.
AV regulation and politics
The experience of AV legislation in Arizona offers two insights. First, regulatory completeness is a mirage, as shifting circumstances, priorities, and contexts require iterative or incremental policy adjustments. Regulatory agility is dependent not only on cross-agency collaboration but also on information feedback into decision-making systems and willingness to take an evolutionary rather than comprehensive approach.
Arizona’s experience with AV regulation reflects this dynamic. In 2015, the governor’s office issued its first executive order on AVs, permitting the testing of AVs on university campuses. Two years later, the state legislature introduced regulations on truck-platooning (electronically assisted caravaning). The governor issued a second executive order concerning the testing and rollout of AVs, holding that AVs must mitigate safety risks and comply with federal safety standards and applicable (including local) traffic laws. Various House Bills followed, addressing issues like personal delivery devices and codifying elements of previously issued executive orders. In 2022, a Senate Bill outlined regulations for Neighborhood Occupantless Electric Vehicles (NOEVs). Arizona’s establishment of an Institute of Automated Mobility (IAM) further strengthened the state’s momentum in AV research and innovation. This experience regulating AVs reflects the type of evolutionary and reactive approach that proponents of light intervention are inclined to support.
As a second insight, public support for such initiatives is dependent on the ability of legislators to leverage existing political dynamics. Pro-growth innovation policy and techno-optimism accord well with popular narratives about free markets. Object lessons in the perils of restrictive regulation have been used to advocate for unfettered markets, particularly in the time since the acceleration of de-regulation in the 1970s. The experience of AVs illustrates how narratives around innovation, technology, and market competition work in combination to promote a policy agenda centered on industry-specific pathways to economic development.
The strongest footing held by opponents of AVs arguably concerns public safety. Without much AV performance data to inform evidence-based design, early-stage policy support may have been considered risky. Over time, however, favorable safety data strengthened the case for AVs. For example, a 2025 study analyzing 56.7 million miles of robotaxi operations shows striking reductions compared to human drivers in traffic incidents at intersections and incidents involving bicyclists and pedestrians.
Other narratives have been deployed to ally the AV phenomenon with various political ambitions. For example, the Goldwater Institute states that AVs present “a unique opportunity to drastically enhance road safety, provide better transportation options for elderly or disabled individuals, reduce traffic, and improve air quality, all while creating vast economic gains from improving efficiency in the shipping industry.” AVs have also been hailed as a safer option for women traveling alone. The promise that AVs will reshape urban transportation, reduce traffic fatalities, and facilitate equitable mobility is durable political leverage.
Recommendation: Evolve or Perish
For governments regulating rapidly evolving technologies, Arizona’s AV experience offers three lessons. First, as I argued in May 2025 at the Midwest Symposium on Social Entrepreneurship in Kansas City, AI policies should balance opportunity with precaution. The stringency of a given regulatory intervention is determined, among other factors, by political appetite for associated tradeoffs (e.g., safety versus rapid innovation). The initial and riskiest stage of the AV learning curve was steep, but has now given way to a more stable and incremental stage of research and information feedback. A 2025 incident near the Phoenix airport, in which an occupied Waymo car repeatedly circled a parking lot, suggests that the learning opportunities are not finished. Early in Arizona’s AV experimental period, the director of the state’s Department of Transportation stated “we shouldn’t be getting in the way by prescribing regulations when we really don’t know how the equipment will perform.” This quote illustrates a distinctly noninterventionist philosophy in how innovation governance is approached amidst uncertainty.
In May 2025, the formative relationship between Arizona and Waymo reached new heights with the announcement of a substantial scale-up of the company’s vehicle manufacturing capacity in the Phoenix area city of Mesa. Katie Hobbs, Arizona’s current Democratic Governor, stated in Waymo’s press release that “the new Waymo and Magna manufacturing facility in Mesa is the latest example of Arizona being the new home for technology to innovate and grow.” The Phoenix area has also seen growth in the construction of data centers and semiconductor manufacturing, highlighting the region’s techno-boosterish ambitions and sending a signal that the local government seeks to match Silicon Valley’s innovative entrepreneurialism. The indelible momentum of the AV industry makes Arizona a useful national test case for policymaking.
Second, policy harmonization is essential in a multi-layered and fragmented governance setting. Absent a federal framework for AVs, states have been largely free to regulate. While the strategy to adopt regulations at a statewide scale has ensured a relatively harmonized governance regime across Phoenix’s numerous municipalities, individual jurisdictions hoping to gain a competitive edge face the challenge of establishing differentiation within regulatory boundaries. The optimal balance between preemption and local policy sovereignty differs across time and issue, and thus has no uniform template.
Accordingly, governance models and technology co-evolve in ways that require continual regulatory recalibration and policy learning. Guidance from federal authorities, such as the Department of Transportation or the National Highway Traffic Safety Administration, could address safety standards while allowing state or local flexibility in implementation. Absent such guidance for the time being, evolution of AV regulation in a context like Arizona’s may reward the jurisdiction with the lowest regulatory ‘common denominator’ – the least intervention allowed by the next highest level of governance.
Third and finally, community input is crucial for building policy legitimacy. Public perceptions about technology are somewhat mercurial, even as AVs themselves (independent of broader technology trends) are not necessarily a politically divisive issue. In Arizona, governors representing both major political parties have championed the industry. However, concerns are growing about equality, safety, and livelihoods. A 2023 study highlights how social inequities have been exacerbated in Arizona’s AV rollout. Concerns persist around privacy and the prospect of hacking or cyber-attacks, and public safety incidents highlight the risks of an AV adhering to a pre-determined route or destination even if circumstances call for evasive or cautionary action. Further, the rapid rollout of robotaxis in China has raised concerns about the hollowing-out of the gig economy and loss of livelihoods.
In San Francisco, AV rollout was more complicated than in Phoenix, with protestors physically obstructing AV movements and testing problems on the city’s crowded streets doing little to alleviate public distrust. Dissonance between ambitious state regulators and more cautious local authorities in San Francisco exemplified the misadventures of coordination failure. This experience shows that the AV phenomenon can become a political flashpoint in an era where stark divisiveness characterizes nearly every policy issue. The distinguishing factor in Arizona’s experience was a more receptive attitude among local governments and prospective users.
In closing, Arizona’s AV regulatory experience highlights broader trends in politics and governance. Assertion of place-based political identities, reflecting a growing urban-rural political divide, may complicate statewide regulatory harmonization around the country. In general, pro-innovation ideas can be seen as economic conservatism, but backlash and skepticism against new technologies and the ‘big tech’ corporations that develop them span the political spectrum. While Arizona’s experience of regulating AVs illustrates the functionality of bipartisan governance, new lessons may be hard learned as innovation continues to evolve more rapidly than policy – and politics – can understand and react.
Kris Hartley is Assistant Professor of Sustainability and Enterprise at the School of Sustainability, Arizona State University.