A moratorium on state laws targeting AI would safeguard innovation and interstate commerce 

In its AI Action Plan, the White House made clear its support for a federal moratorium on new state and local laws regulating artificial intelligence (AI). A federal moratorium on state and local laws and regulations governing the design and performance of AI models is justified to preserve innovation and interstate commerce, in keeping with the basic principles of federalism in the United States. 

Many states have passed or are considering multiple AI bills that are technically complex and politically charged, singling out this technology for extra regulation on issues like discrimination, child safety, and housing. This avalanche of legislation is especially troubling because existing laws in all these areas apply equally to AI, and the current need to single out the technology is highly dubious.  

In making the case for a federal moratorium, many observers correctly point to the danger of a “regulatory patchwork” in which up to 50 states may make widely varying rules and stifle important and useful innovations. There is no natural reason for data to stop at state borders, meaning virtually all economic activity involving AI is interstate commerce. The framers of the U.S. Constitution recognized interstate commerce as essential for both economic prosperity and political unity, placing its control in federal hands. In addition to reducing inefficiencies arising from inconsistent state laws, federal powers over interstate commerce using AI prevent large states such as California from gaining the de facto power to regulate the new technology across the entire country. 

AI is a foundational technology with an impact many expect to be on par with the internet or personal computers. Like those technologies, AI is already spawning an abundance of applications for a wide range of uses. This makes the development of AI extremely important, but also highly uncertain. There is no single finish line in the “AI race.” While building the technology and its countless applications will require massive efforts, they cannot be coordinated toward a single goal. This fundamental uncertainty is the source of AI’s great potential. AI breakthroughs will be made not only in university laboratories, but by firms, workers, and entrepreneurs in our market economy. This process of innovation will ultimately lead us toward finish lines we can’t yet even imagine. 

Nothing about AI places the technology or its applications outside the reach of any existing laws or government agencies with respect to any policy. Just as we remain early in the process of finding out what AI can do, we have much to learn before concluding that AI requires different laws or approaches to regulation. But proponents of restrictive AI legislation today are guilty of exactly this mistake.  

State bills seeking to govern the design and performance of AI systems are not typically written with today’s chatbots in mind. They are concerned about what future AI technology might do, or what our future energy and data center needs might be. Supporters of such efforts do not appreciate that transformative technologies and the societies they transform change together. Those who worry about AI replacing workers, for example, usually miss that such technological changes happen over time rather than all at once. Workers, therefore, can and do adjust to new technology. 

As the path of AI’s development becomes clear over the coming years and decades, speculative and restrictive laws will do more harm than if they were simply outdated rules that missed the mark. Government regulation influences the course taken by technology. When that course is based on flawed assumptions, we fail to realize the benefits of technology fully and often don’t know what those benefits might have been. 

Compounding these difficulties and potential for misunderstanding is the expansion in the list of technologies commonly called “AI.” Once associated with science fiction and futurism, the term “AI” gained widespread use in the last few years, after generative AI chatbots powered by large language models were introduced. However, people now frequently apply the term to many algorithmic tools widely used for years or even decades. 

There is no current need to single out AI, over and above other technologies or means of expression, for regulation or enforcement beyond existing laws. Nonetheless, the political temptation of singling out real or imagined use of AI is too great for many to resist. Users of AI technology in housing and labor markets are singled out for more oversight and compliance under anti-discrimination law. Large AI developers are singled out for more transparency rules, seemingly at the discretion of state officials. Images of political figures created using AI are singled out and banned in the lead-up to elections. AI-powered human resources software suites face extra scrutiny. 

The correct approach toward AI at all levels of government is to enforce existing laws and regulations. Indeed, Congress’ AI moratorium proposal from earlier this year explicitly exempted from federal preemption “generally applicable” state and local laws and regulations. Comprehensive legislation targeting AI may prove desirable in the future, but that day has not arrived. While some might consider such a level-headed approach from the federal government unlikely, expecting similar restraint in 50 statehouses at once is all but impossible. Absent a moratorium, state legislators would nevertheless be well advised to exercise and advocate such restraint. 

State lawmakers can also pass legislation that affirms the applicability of current laws to AI and establishes basic individual freedoms with respect to computing. Bills such as Right to Compute, passed in Montana and introduced in an increasing number of states, provide an important blueprint for those who see the importance of restraint to make that case and partially mitigate the efforts of overactive colleagues. 

A torrent of bills singling out AI in misguided attempts to solve what amount to guesses of future problems would threaten the open climate of innovation and entrepreneurship that has greatly benefited the United States in recent decades. But uncertainty, misunderstanding, political incentives, and media hype conspire in the current moment to produce this outcome in statehouses across the country. A federal moratorium on bills singling out AI would help ensure the U.S. remains fertile ground for the AI revolution. 

The post A moratorium on state laws targeting AI would safeguard innovation and interstate commerce  appeared first on Reason Foundation.