Jennifer Huddleston
In 2023, many “doomers” who worried about artificial intelligence (AI) growth called for a moratorium on AI development in a viral letter. In the two years since, however, not only has the doomsday scenario failed to occur, but AI has both advanced and become more accepted by consumers for its usefulness and entertaining applications, such as turning your pet’s photo into what it would look like as a human.
Pausing the Patchwork
Not surprisingly, the idea of a ten-year state moratorium, included in the Big Beautiful Bill, has sparked some panic of its own.
This criticism has come from both sides of the aisle. On the right, the right to regulate may be valued highly due to strong beliefs in federalism and local control when possible. For example, countless Republican state attorneys general have joined a bipartisan letter opposing the legislation on the grounds that it is not “respectful to states.” On the left, the criticism comes more from a belief that the technology needs regulation by someone and an appreciation for the more European-style regulatory approach.
States have already begun to consider and pass a significant amount of AI-related legislation, raising an increased risk that a patchwork of rules could emerge. These laws risk, at best, adding a significant cumulative amount of regulation that could deter a technology and industry in its early phase or, at worst, conflicting with one another where products cannot operate in certain jurisdictions. This includes not only the enjoyable consumer products like ChatGPT or Google’s Gemini but could impact the important benefits in areas like medicine, natural disaster response, and more.
This patchwork would also burden US innovators by forcing them to determine the regulatory regime of fifty different states and focus on compliance rather than innovation. This is not a smart policy approach, especially at a time when industry leadership remains undetermined with strong competition from places like China.
In its current form, states can either choose to reject BEAD (Broadband Equity, Access, and Deployment) funding and continue to regulate AI or be subject to the moratorium. This approach means it is still possible that a state patchwork could emerge; however, it signals an important shift in the Overton window around the conversation on regulating AI. Ideally, this is not a technology to rush to regulate out of fear of the worst-case scenario, but rather something that has significant benefits to carefully consider in a thoughtful manner, not a rushed one.
Building a Federal Framework for Innovation
A moratorium on state-level policy could provide the impetus for Congress to identify and determine where such a framework is needed while still allowing innovation to flourish without Sacramento or Austin disrupting progress for everyone.
Notably, the moratorium only applies to interstate actions, meaning states would still likely be able to determine their intrastate requirements or regulations, such as around procurement and law enforcement restrictions. Similarly, it is reasonable to presume that the law does not prevent tech-agnostic regulation updates that happen to apply to AI as well, such as updating fraud statutes, but does prevent changing laws to single out the particular technology.
In many ways, consumer AI products are roughly akin to the Internet in the mid- to late 1990s, and this policy approach would set up a continuation of that philosophy. In that time, the US chose to take a light-touch approach that allowed innovators and entrepreneurs, not government bureaucrats, to determine the best ways to utilize the new World Wide Web. This did require a policy framework that included pre-empting state regulation in some cases, like Section 230 and the Internet Tax Freedom Act, lest states create laws that could deter or disrupt this new technology.
Of course, more recently, some concerning patchworks impacting the internet around data privacy have emerged, but these early internet laws limited a single state, causing disruption. Hopefully, this signals that policymakers are recognizing why such an approach was important for the development of the internet and its applications as it considers this new leap in general-purpose technology nearly 30 years later.
Conclusion
While to some a moratorium on state AI policy may seem like a crude tool or federal overreach, the reality is that the nature of this technology is almost always naturally interstate, making it a federal, not a state, issue. There are positive actions states can take, such as interpreting and updating existing regulations and considering civil liberties protections from their governments’ use of AI. In many cases, such actions would be intrastate and relatively unaffected.
Already, many of the “doomers’” fears have turned out to be hyperbolic and unlikely. A decade gives time for both societal norms and a federal framework to adapt to this new technology without the fear that one state’s regulation may derail the important and beneficial progress.
