Introduction
The rapid proliferation of artificial intelligence (AI) across all sectors of society has presented a significant regulatory challenge for lawmakers in the United States. As AI systems become more integrated into critical areas such as employment, finance, healthcare, and law enforcement, questions surrounding accountability, fairness, privacy, and safety have become increasingly urgent. In the US federal system, this has ignited a debate over the appropriate level of government at which to legislate. The core of this debate concerns whether a single, comprehensive federal AI law should pre-empt, or override, the growing number of AI-related laws emerging from individual states. Proponents of federal pre-emption argue for national uniformity to foster innovation and avoid a confusing patchwork of regulations. Conversely, opponents fear that a uniform federal law would stifle state-level democratic experimentation and potentially weaken consumer protections. This essay will argue that while the pursuit of national uniformity has considerable merit, a pre-emptive federal AI law should only be enacted once it can guarantee robust and comprehensive standards. To pre-empt emerging state laws without a strong federal alternative would prematurely halt valuable regulatory innovation and risk establishing a weak national standard that prioritises business interests over public protection.
The Argument for Federal Pre-emption
The primary argument for federal pre-emption of state AI laws is rooted in the desire for a coherent and uniform national regulatory landscape. This position, often advanced by technology companies and business coalitions, is based on principles of economic efficiency, legal certainty, and international competitiveness. The legal authority for such action derives from the Supremacy Clause of the U.S. Constitution, which establishes that federal laws made pursuant to the Constitution shall be the "supreme Law of the Land" (U.S. Const. art. VI, cl. 2). Where Congress acts to regulate an area, it can expressly or implicitly pre-empt state laws on the same subject.
One of the most compelling arguments for pre-emption is the avoidance of a regulatory ‘patchwork’. If each of the 50 states develops its own unique rules for AI, companies that operate nationally would face enormous compliance burdens. A business developing an AI-powered service would need to navigate potentially conflicting requirements regarding data usage, algorithmic transparency, impact assessments, and consumer rights in every state in which it operates. As the U.S. Chamber of Commerce (2023) has argued, this fragmentation could create significant legal uncertainty, increase costs, and ultimately stifle innovation by making it difficult for developers, particularly smaller businesses, to scale their products. A single federal standard would, in theory, create a level playing field and simplify compliance, allowing companies to focus resources on developing new technologies rather than on navigating a maze of state-specific regulations.
Furthermore, a unified federal approach is seen as essential for maintaining the United States' position as a global leader in AI development. Proponents of pre-emption contend that a fragmented domestic market puts American companies at a disadvantage against competitors from jurisdictions with cohesive national strategies, such as China or the European Union with its comprehensive AI Act (European Commission, 2021). A strong, clear federal framework could signal stability to international markets and partners, promoting investment and ensuring that the US plays a leading role in setting global norms for responsible AI. It could also be argued that certain AI applications, particularly those related to national security and critical infrastructure, demand a centralised federal response that cannot be adequately handled by individual states.
Finally, supporters of pre-emption argue that a federal law could provide a consistent baseline of consumer protection for all Americans. Rather than rights varying depending on one's state of residence, a federal law would guarantee a uniform set of protections nationwide. However, this argument is heavily dependent on the substance of the federal law itself, a point of significant contention. The ideal federal law, from this perspective, would be one that is both comprehensive enough to be meaningful and flexible enough to adapt to technological change, thereby creating a stable and predictable environment for both consumers and industry.
The Case for State-Led Regulation and Experimentation
In direct opposition to the push for pre-emption is the view that states should retain the power to regulate AI within their borders. This perspective draws on the long-standing American constitutional tradition of federalism and the concept of states as "laboratories of democracy." This idea, articulated by Justice Louis Brandeis, suggests that "a single courageous State may, if its citizens choose, serve as a laboratory; and try novel social and economic experiments without risk to the rest of the country" (New State Ice Co. v. Liebmann, 1932, p. 311). In the context of a rapidly evolving technology like AI, allowing states to innovate in their regulatory approaches is seen as a vital mechanism for developing effective and contextually appropriate rules.
States are often more nimble and responsive to the specific concerns of their populations than the federal government. For example, Illinois enacted the Biometric Information Privacy Act (BIPA) in 2008, long before facial recognition and other biometric technologies became a major federal issue. This law has provided Illinois residents with strong protections and a private right of action, leading to significant settlements and influencing how companies handle biometric data nationwide (Illinois General Assembly, 2008). Similarly, New York City’s Local Law 144 specifically targets the use of automated tools in employment decisions, requiring bias audits and transparency for job applicants (City of New York, 2021). These examples show states identifying and addressing specific harms in a targeted way, often years before a consensus might emerge at the federal level. Pre-empting such laws could eliminate tailored solutions that address pressing local problems.
Moreover, states have historically served as the vanguard of consumer protection. The California Consumer Privacy Act (CCPA), later expanded by the California Privacy Rights Act (CPRA), established a new benchmark for data privacy in the US and effectively forced companies to grant new rights to consumers (State of California, 2018). The success of California's legislation has inspired several other states, such as Colorado and Virginia, to pass their own comprehensive privacy laws. Consumer advocacy groups argue that this state-level momentum is crucial. They fear that federal pre-emption would result in a "race to the bottom," whereby industry lobbyists would pressure a gridlocked Congress into passing a weak federal law that overrides stronger state protections. In this scenario, pre-emption would turn the protective "floor" that a federal law could provide into a restrictive "ceiling," preventing states from offering greater protections to their citizens (Kaminski, 2023). This concern is central to the resistance from state attorneys general and consumer groups, who argue that Congress should not wipe out state laws without first creating a robust and comprehensive federal replacement (NAG, 2023).
Evaluating the Path Forward: Pre-emption with a High Floor
The debate between national uniformity and state experimentation is not a simple binary choice. The optimal path forward likely involves a synthesis of both federal and state power. The central question should not be whether the federal government should act, but how it should act. The most significant danger identified by critics of pre-emption is the prospect of a weak federal bill that eliminates stronger state laws. Therefore, any discussion of pre-emption must be intrinsically linked to the substance of the proposed federal legislation.
A potential compromise can be found in the model of "floor pre-emption" rather than "ceiling pre-emption." Under this model, a federal AI law would establish a mandatory baseline of rights and protections that applies across the entire country. This would address the industry's desire for a level playing field and eliminate the most permissive state-level approaches. However, it would explicitly preserve the authority of states to enact stronger or more specific laws to supplement the federal standards. This approach has precedent in other areas of US law, such as the Clean Air Act, which sets federal air quality standards but allows states like California to implement stricter emissions controls (U.S. Environmental Protection Agency, n.d.). Such a framework would allow the federal government to ensure a foundational layer of protection for all citizens while still empowering states to act as laboratories of democracy, addressing unique local concerns or pioneering more advanced safeguards.
However, the political reality is that passing any major legislation through a divided Congress is challenging. The technology industry wields considerable influence, and there is a real risk that any pre-emption clause would be part of a legislative package that favours business interests over consumer protection. Given this context, the argument that a push for pre-emption is currently premature holds significant weight. As stated by a coalition of civil rights and consumer groups, pre-emption should be off the table "unless and until Congress has created a new federal agency and passed a law that provides robust, meaningful protections" (The Leadership Conference on Civil and Human Rights, 2023). Until such a framework exists, state-level laws, despite creating a complex compliance environment, serve as the only meaningful source of regulation for AI harms. They provide immediate protections for consumers and create valuable case law and regulatory experience that can, and should, inform the eventual development of a stronger federal bill.
Conclusion
The question of whether federal AI legislation should pre-empt state laws lies at the heart of the American system of federalism, balancing the need for national consistency against the value of local autonomy. The arguments for a uniform federal standard are compelling, particularly regarding economic efficiency and the avoidance of a burdensome regulatory patchwork for businesses operating nationwide. A single set of rules could foster innovation and strengthen the US position on the global stage.
However, these benefits would come at too high a cost if national uniformity is achieved by sacrificing meaningful consumer protection and democratic experimentation. States have proven to be agile and effective regulators, often stepping in to protect citizens from emerging technological harms far more quickly than the federal government. To erase these state-level protections with a weak or industry-friendly federal law would be a profound setback for public interest.
Therefore, this essay concludes that federal pre-emption of state AI laws is, at present, an inadvisable course of action. The risk of establishing a low regulatory ceiling is too great. The "laboratories of democracy" should be allowed to continue their work, developing novel approaches to AI governance from which the entire nation can learn. The ideal future involves federal legislation, but it must be one that sets a high protective floor, not a restrictive ceiling. Until Congress is prepared to pass a truly robust, comprehensive, and well-enforced AI law, the varied and sometimes complicated landscape of state regulation remains the most effective tool for ensuring that the development of AI serves the public good.
References
City of New York. (2021) Local Law 144 of 2021. Available at: https://legistar.council.nyc.gov/LegislationDetail.aspx?ID=3883325&GUID=F4200366-7287-4351-893A-2475E2333E3E (Accessed: 15 May 2024).
European Commission. (2021) Proposal for a Regulation laying down harmonised rules on artificial intelligence (Artificial Intelligence Act). Available at: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A52021PC0206 (Accessed: 15 May 2024).
Illinois General Assembly. (2008) 740 ILCS 14/ Biometric Information Privacy Act. Available at: https://www.ilga.gov/legislation/ilcs/ilcs3.asp?ActID=3004&ChapterID=57 (Accessed: 15 May 2024).
Kaminski, M. E. (2023) 'The States Will Lead on AI Regulation', Lawfare. 31 July. Available at: https://www.lawfaremedia.org/article/the-states-will-lead-on-ai-regulation (Accessed: 15 May 2024).
National Association of Attorneys General (NAG). (2023) Bipartisan Coalition of 51 Attorneys General Urges Congress to Regulate Artificial Intelligence. Available at: https://www.naag.org/press-releases/bipartisan-coalition-of-51-attorneys-general-urges-congress-to-regulate-artificial-intelligence/ (Accessed: 15 May 2024).
New State Ice Co. v. Liebmann, 285 U.S. 262 (1932).
State of California Department of Justice. (2018) California Consumer Privacy Act (CCPA). Available at: https://oag.ca.gov/privacy/ccpa (Accessed: 15 May 2024).
The Leadership Conference on Civil and Human Rights. (2023) Coalition Letter to Congress on Artificial Intelligence Policy. Available at: https://civilrights.org/resource/coalition-letter-to-congress-on-artificial-intelligence-policy/ (Accessed: 15 May 2024).
U.S. Chamber of Commerce. (2023) Artificial Intelligence Commission Report. Washington, D.C.: U.S. Chamber of Commerce.
U.S. Constitution, Art. VI, cl. 2.
U.S. Environmental Protection Agency. (n.d.) Clean Air Act Requirements and History. Available at: https://www.epa.gov/clean-air-act-overview/clean-air-act-requirements-and-history (Accessed: 15 May 2024).
