AI Regulation in the United States

AI regulation in the United States is a patchwork of executive orders, federal agency guidance, procurement rules and state legislation, rather than a single comprehensive AI statute of the kind adopted by the European Union. Federal policy has swung sharply between administrations — from the safety-and-rights agenda of the 2023 Biden executive order to the Trump administration's doctrine of accelerating AI development and asserting global dominance — leaving businesses to navigate a rapidly changing compliance landscape.^[1][2]

History

The modern federal phase began with a 2019 executive order establishing the American AI Initiative, which directed agencies to prioritise AI research and set the tone for a research-and-competitiveness framing. In January 2023 the National Institute of Standards and Technology published the AI Risk Management Framework, a voluntary governance standard that remains widely used by organisations even after later policy changes. Sectoral regulators — among them financial agencies, health regulators and the Federal Trade Commission — continued to apply existing law to AI applications.^[2][3]

On 30 October 2023, President Joe Biden signed Executive Order 14110, "Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence". It imposed safety testing and reporting obligations on the largest frontier models, required agencies to assess AI risks across areas including civil rights and critical infrastructure, and launched a wave of implementation guidance. It was also controversial with parts of the technology industry, which argued it slowed innovation.^[3]

The 2025 change of administration brought rapid reversal. On 20 January 2025, Executive Order 14148 revoked EO 14110, and on 23 January 2025 Executive Order 14179, "Removing Barriers to American Leadership in Artificial Intelligence", established the official policy of sustaining and enhancing America's AI dominance and directed the development of a new action plan. April 2025 brought revised Office of Management and Budget guidance on federal agencies' own use and procurement of AI.^[1][2]

On 23 July 2025 the White House released America's AI Action Plan, organised around three pillars — accelerating innovation, building AI infrastructure, and international diplomacy and security — and accompanied by executive orders on federal procurement and ideological neutrality in AI ("Preventing Woke AI in the Federal Government"), on accelerating federal permitting for data-centre infrastructure, and on promoting the export of the American AI technology stack. Later in 2025, further orders directed AI to be applied to scientific discovery through the Genesis Mission (November, with more than US$5 billion announced for the programme by mid-2026) and to specific efforts such as paediatric cancer research.^[4][5]

On 11 December 2025, an executive order titled "Ensuring a National Policy Framework for AI" moved to centralise policy by directing the Department of Justice to challenge burdensome state AI laws, conditioning certain federal funding on policy alignment and pressing Congress for a uniform federal framework. The order responded to what the administration described as a fracturing landscape — more than 1,000 AI-related bills were introduced in state legislatures during 2025 — and continued in 2026 with further executive actions on AI innovation, security and national security applications.^[6][7]

Key Concepts

Governance by executive order. Because Congress has not enacted a comprehensive AI statute — the only significant AI-adjacent federal law of the period, the TAKE IT DOWN Act on non-consensual intimate imagery, was enacted in May 2025 — federal AI policy has been made and unmade by executive action. This makes it unusually volatile: the 2025 reversal of EO 14110 demonstrated that a change of administration can reset entire compliance programmes.^[6][7]

Federalism and preemption. The United States is simultaneously one of the most active and most contested regulatory arenas for AI, with individual states — including Colorado, California and Texas — advancing their own AI statutes covering discrimination, transparency and high-risk use, while the federal executive has sought to preempt them. For companies operating nationally, the resulting uncertainty is itself a compliance burden.^[6]

Voluntary standards alongside hard law. Instruments such as the NIST AI Risk Management Framework remain the practical compliance backbone for many firms, complementing sectoral enforcement and export controls on advanced semiconductors — which in turn shape global access to AI computing power.^[2][3]

Applications and Impact

United States policy choices have global reach: American model developers dominate the frontier of AI capability, and their home-country rules — or the absence of them — influence how their products are governed everywhere else. The export-promotion agenda sought to spread American AI systems and chips to allied countries, while European regulation such as the EU AI Act applies to many of the same companies, creating overlapping obligations.^[4][6]

Within the United States, the policy swing intensified debate between innovation-first and safety-first camps: proponents of the 2025 direction argue that lighter regulation protects investment and competitiveness against China, while critics warn of weak safeguards for consumers, workers and civil rights, and of regulatory uncertainty created by preemption fights. The trajectory of United States policy is closely followed by governments — including Malaysia's — that must align domestic governance frameworks with a global AI market shaped largely by American firms.^[6]

>See Also

• EU AI Act
• AI Regulation Malaysia
• National AI Office Malaysia
• ASEAN AI Governance
• PDPA AI Compliance