Guiding Principles for State AI Policy

Download
Tanner Jones

Tanner H. Jones

Issue Areas

Related Content

Introduction

As artificial intelligence (AI) becomes integral to how our economy works — transforming productivity and augmenting how businesses produce goods and services for customers at every level—state lawmakers face both opportunities and challenges. While federal oversight risks overreach that could hinder American competitiveness, a fragmented patchwork of state regulations is also worrisome and would block and deter innovation by forcing entrepreneurs to navigate divergent state codes, create new avenues for activist harassment and bureaucratic capture, and impose unnecessary costs on businesses and consumers.

State lawmakers should carefully evaluate whether new legislation is truly needed when addressing technological innovation. In many cases, existing legal frameworks already provide adequate protection against potential harms from emerging technologies. Rather than rushing to create new regulations, legislators should focus on educating their constituents about how current laws can effectively address these challenges. When new policies are necessary, they should be precisely targeted to address specific harmful outcomes while continuing to foster innovation.

Lawmakers must understand that the progressive regulatory framework—premised on rule by unelected “experts” in the executive branch — has proven both ineffective at solving problems and hostile to innovation. Robust economic theory and empirical findings demonstrate that new regulatory powers, even those crafted with the best intentions, are captured by incumbent industry players and legal compliance and consulting firms, all of whom benefit from red tape and barriers to market entry.[i] President Trump and Elon Musk’s DOGE effort federally offers a rebuttal against these ideas, demonstrating that legislators can address bad outcomes without inviting new regulation that crushes innovation and dynamism.

If innovation can thrive, AI promises to revolutionize healthcare through early disease detection and personalized treatment protocols, unlock unprecedented productivity gains across industries and in federal and state government, and cement U.S. hegemony through advanced national security applications.

With several AI-related proposals circulating in state legislatures, it is critical that any new policies align with core principles. The following framework provides guidance for state lawmakers seeking to harness AI’s benefits while mitigating potential risks.[ii]

Five Guiding Principles for State AI Policy

1. Promote Government Efficiency and Model Best Practices

States should lead by example in AI adoption, establishing best practices that the private sector can emulate:[iii]

  1. Leverage AI to remove waste and administrative burdens while improving service delivery
  2. Implement pilot programs with appropriate transparency measures
  3. Document and share lessons learned to help scale successful implementations
  4. Develop clear, fair, and transparent procurement guidelines

2. Nurture Innovation by Removing Regulatory Burdens

States should ensure regulatory and compliance codes do not hinder nascent technologies:[iv]

  1. Legislators and regulators should avoid unnecessary interventions and exercise humility when confronting evolving technology[v]
  2. Create regulatory sandboxes and targeted exemptions from existing regulations for AI startups
  3. Codify mechanisms like automatic sunsets to delete onerous codes and regulations
  4. Repeal archaic frameworks that do not reflect technological advances

3. Exercise Legislative Powers, Do Not Defer to Unelected Regulators[vi]

Do the hard work of lawmaking, do not outsource to unelected officials in the executive branch:

  1. Identify specific gaps in existing criminal, civil, and product liability codes
  2. Draft targeted legislation to address AI applications within established legal frameworks[vii]
  3. Avoid expansive delegations of rulemaking authority
  4. Do not allow agencies to bypass standard promulgation with methods like guidance documents or emergency rules

4. Address Harms Through Outcome-Based Enforcement

Rather than implement preemptive restrictions based on theoretical capabilities, states should:

  1. Leverage existing legal frameworks including tort law and consumer protection statutes
  2. Ensure meaningful legal recourse only when actual injuries occur
  3. Do not constrain nascent technologies by focusing on potential outcomes and capabilities[viii]
  4. Close gaps in certain existing statutes—child sexual images, revenge porn, etc.—to ensure malicious users of generative AI tools are punished with the full force of the law

5. Guard Against Regulatory Capture[ix]

To prevent established players from using regulation to preserve market advantages:

  1. Apply an adversarial mindset to proposed legislation to examine potential for misuse—a concept known as ‘red teaming’[x]
  2. Be aware that many advocacy groups and their backers are pursuing specific and potentially hidden agendas
  3. Resist pressure for unnecessary regulations that primarily benefit compliance industries
  4. Maintain focus on nurturing competition and enabling new market entrants

Implementation Guidance

When evaluating AI-related proposals, state lawmakers should consider:

  • Whether the policy aligns with all core principles
  • Potential to create regulatory fragmentation
  • Ramifications for innovation and market competition
  • Ability to work within existing legal and regulatory frameworks
  • Administrative feasibility and enforcement
  • Flexibility to adapt to technological change

Examples of sound approaches:

  • Exempting AI startups from existing regulations to nurture innovation
  • Amending child porn laws to ensure bad actors using AI to generate these images are prosecuted and sentenced
  • Using AI tools to expedite the delivery of government services, reduce the size of government bureaucracy, and save tax dollars

Examples of counterproductive approaches:

  • Requiring licenses for basic AI implementations
  • Mandating extensive pre-deployment testing requirements
  • Creating new bureaucracies to oversee AI development

Conclusion

State AI policy should balance innovation with targeted safeguards by leveraging existing legal frameworks where possible, resisting the urge to delegate new regulatory authority (and create new regulators), and be vigilant against capture by special interests. Success requires lawmakers to lead by example in AI adoption while maintaining legislative control rather than deferring to unelected regulators. Cicero’s approach will help states harness AI’s transformative potential across healthcare, productivity, and national security while addressing specific harms through precise, outcome-based enforcement.

[i] Stigler, George J. 1971. “The Theory of Economic Regulation.” The Bell Journal of Economics and Management Science 2 (1): 3–21. https://doi.org/10.2307/3003160.

[iii] Longe, Edward. 2024. “Bringing Government into the 21st Century: Artificial Intelligence and State Government Operations – James Madison Institute.” James Madison Institute. May 2024. https://jamesmadison.org/jmi-releases-new-policy-brief-on-artificial-intelligence-and-state-government-operations/.

[iv] “Principles for Lawmakers: How to Think about Emerging Technologies.” n.d. Accessed January 24, 2025. https://pelicanpolicy.org/wp-content/uploads/2024/04/AI-Toolkit_Principles-For-Lawmakers_How-To-Think-About-Emerging-Technologies.pdf.

[v] Pai, Hao-Kai. 2017. “Regulatory Humility Humility.” American Enterprise Institute – AEI. October 25, 2017. https://www.aei.org/technology-and-innovation/regulatory-humility-humility/.

[vi] “Rules for Robots – a Framework for Governance of AI.” 2023. Competitive Enterprise Institute. October 19, 2023. https://cei.org/studies/rules-for-robots-a-framework-for-governance-of-ai/.

[vii] Erdélyi, Olivia J., and Judy Goldsmith. 2018. “Regulating Artificial Intelligence.” Proceedings of the 2018 AAAI/ACM Conference on AI, Ethics, and Society, February. https://doi.org/10.1145/3278721.3278731.

[viii] “Regulating Machine Learning Open-Source Software.” 2024. Abundance.institute. 2024. https://abundance.institute/articles/regulating-machine-learning-open-source-software.

[ix] Stigler, George J. 1971. “The Theory of Economic Regulation.” The Bell Journal of Economics and Management Science 2 (1): 3–21. https://doi.org/10.2307/3003160.

[x] Chilson, Neil. 2024. “Red Teaming AI Legislation: Lessons from SB 1047.” Substack.com. Getting Out of Control. August 25, 2024. https://outofcontrol.substack.com/p/red-teaming-ai-legislation-lessons.

Stay Informed

Sign up to receive updates about our fight for policies at the state level that restore liberty, accountability, and innovation in American governance.