Loading
AI regulation is the body of binding law, executive action, and enforceable regulatory guidance that governs how artificial intelligence systems are developed, sold, and used. There is no single global regime: the European Union has enacted comprehensive, risk-tiered horizontal law in the EU AI Act; the United States relies on existing sectoral regulators layered under a fast-moving federal-state contest over who sets the rules; and jurisdictions such as the UK, Australia, Singapore, and Japan favour principles-based regulator guidance over hard law. The landscape shifted materially on 11 December 2025, when the White House signed Executive Order 14365, "Ensuring a National Policy Framework for Artificial Intelligence," directing federal agencies to identify, challenge, and potentially defund state AI laws it views as burdensome, a move now being actively contested by Congress and the states. For governance and compliance teams, AI regulation is not a static reference: which law actually applies to a given AI system, in a given jurisdiction, is genuinely contested right now and can shift within months rather than years.
Run the free AI Health CheckAI Regulation, the body of binding law and enforceable regulatory guidance that governs how AI systems are developed, sold, and used.
The global AI regulatory landscape spans three broad approaches: the EU AI Act's comprehensive horizontal regulation; principles-based regulator guidance over hard law in Australia, the UK, Singapore, and Japan; and the US's traditional sectoral approach, in which existing regulators (FTC, EEOC, SEC, OCC, FDA, CFPB) apply existing law to AI alongside a growing state law patchwork. That US model is now contested: Executive Order 14365 (11 December 2025) directs a DOJ AI Litigation Task Force, stood up 9 January 2026, to challenge state AI laws and ties federal broadband funding to compliance, seeking to preempt the state patchwork rather than merely coexist with it, though no state law had been struck down as of mid-2026 and Congress has introduced bills, such as the GUARDRAILS Act, to nullify the order. Frontier AI is increasingly governed by voluntary commitments to safety institutes: the US Center for AI Standards and Innovation (CAISI, renamed from the US AI Safety Institute in June 2025 with its mandate shifted from safety-first toward industry testing, security evaluation, and international standards competitiveness), UK AISI, and Australian AISI.
Source: EU AI Act; Executive Order 14365 (2025); OECD AI Policy Observatory
European Union: comprehensive horizontal law
Regulation (EU) 2024/1689 (the EU AI Act) classifies AI by risk tier, prohibited, high-risk, limited-risk, minimal-risk, and applies extraterritorially to any AI system placed on the EU market or whose outputs affect people in the EU. Obligations phase in from February 2025 through 2027, with several deadlines revised by the June 2026 Digital Omnibus, adopted by the Council on 29 June 2026.
United States: sectoral regulation plus an escalating federal-state fight
No single federal AI statute exists. Existing regulators, the FTC, EEOC, SEC, OCC, FDA, and CFPB, apply existing law to AI case by case, while states have filled the gap with their own laws. Since December 2025, a White House executive order has directed the federal government to challenge and defund state AI laws it considers inconsistent with a 'minimally burdensome' national policy, an active and unresolved conflict rather than a settled model.
Principles-based regulator guidance: UK, Australia, Singapore, Japan
These jurisdictions govern AI mainly through regulator guidance, voluntary frameworks, and existing cross-sector law (privacy, consumer protection, anti-discrimination) rather than a dedicated AI statute. Examples include Singapore's MAS FEAT principles and AI Verify toolkit, and sector-regulator-led guidance in the UK and Australia.
Frontier AI: voluntary commitments to national institutes
The largest AI labs make voluntary safety and security commitments to national AI institutes rather than a binding frontier-AI statute. In the US this is now the Center for AI Standards and Innovation (CAISI), renamed from the US AI Safety Institute in June 2025 with a mandate shifted toward industry testing and standards competitiveness; the UK and Australia run comparable institutes.
On 11 December 2025, the White House signed Executive Order 14365, 'Ensuring a National Policy Framework for Artificial Intelligence.' It declares it US policy to sustain American AI leadership through a single, minimally burdensome national framework rather than a state-by-state patchwork, and it directs several federal levers to make that happen. Within 30 days, the Attorney General was to establish an AI Litigation Task Force with the sole job of challenging state AI laws in court on grounds including unconstitutional interstate-commerce regulation and federal preemption; Attorney General Bondi formally stood up the Task Force on 9 January 2026.
The order also directed the Secretary of Commerce to publish, within 90 days, an evaluation identifying state AI laws considered 'onerous' and suitable for referral to the Task Force, and it directed agencies to assess whether federal broadband funding under the BEAD program should be withheld from states whose AI laws conflict with the administration's policy. The order explicitly singled out Colorado's since-repealed AI Act as an example of the kind of state law it targets.
Congress pushed back quickly. Senator Ed Markey introduced the States' Right to Regulate AI Act (S.3557) in the Senate just six days after the order, on 17 December 2025, to bar federal funds from being used to implement it. Representative Don Beyer and colleagues followed with the House GUARDRAILS Act (H.R. 8031) on 20 March 2026, and Senator Brian Schatz introduced a Senate companion (S.4216) shortly after; each of these bills would declare the executive order to have no force or effect and bar federal funds from being used to implement it. None of these bills had passed both chambers as of mid-2026.
As of July 2026, no state AI law has been struck down or successfully preempted through this mechanism. Legal trackers reported the AI Litigation Task Force had not yet filed a lawsuit, with commentary anticipating its first case, widely expected to target Colorado, later in 2026. In practice, this means state AI laws remain enforceable unless and until a court rules against them or a state repeals its own statute (as Colorado itself did in May 2026, replacing its original AI Act with a narrower, disclosure-based law under SB 189). Treating the executive order as having already removed state-level compliance obligations would be premature.
2 February 2025
EU AI Act Article 5 prohibited practices take effect, the first EU AI Act obligations to bind.
June 2025
US Department of Commerce renames the US AI Safety Institute to the Center for AI Standards and Innovation (CAISI).
2 August 2025
EU AI Act General-Purpose AI (GPAI) provider obligations take effect.
11 December 2025
Executive Order 14365 signed, ordering a federal push to preempt state AI laws via litigation, funding conditions, and agency review.
17 December 2025
Senator Ed Markey introduces the States' Right to Regulate AI Act (S.3557) in the Senate, just six days after EO 14365, to bar federal funds from being used to implement it.
9 January 2026
US Attorney General formally establishes the AI Litigation Task Force required by EO 14365.
14 January 2026
Taiwan's Artificial Intelligence Basic Act enters into force.
20 March 2026
Representative Don Beyer and colleagues introduce the House GUARDRAILS Act (H.R. 8031) to nullify EO 14365; Senator Brian Schatz introduces a Senate companion (S.4216) shortly after.
14 May 2026
Colorado repeals its original AI Act and replaces it with the narrower, disclosure-based SB 189, effective 1 January 2027.
The practical takeaway for organisations deploying AI is that the US regulatory map is currently in motion at the federal level while state obligations remain legally live. A compliance program built only around 'wait for federal preemption' risks being wrong twice: once if a targeted state law is repealed or struck down, and again if it survives litigation or Congress nullifies the executive order first. The safer posture is to keep tracking each applicable state law on its own merits, note where an executive order or pending litigation could change that law's status, and revisit the assessment as the AI Litigation Task Force's first cases are filed and decided.
More broadly, this is a reminder that AI regulation is not a fixed reference document anywhere in the world right now. The EU AI Act's own deadlines have already been renegotiated once, via the June 2026 Digital Omnibus, adopted by the Council on 29 June 2026; the US federal-state conflict could resolve through court rulings, congressional action, or a change in administration priorities; and jurisdictions using principles-based guidance can convert that guidance into hard law with little notice. Organisations with cross-border AI operations need a process for monitoring regulatory change, not a one-time compliance mapping exercise.
Is there a comprehensive federal AI law in the United States?
No. As of July 2026, Congress has not passed a comprehensive federal AI statute. The US instead relies on existing sectoral regulators (FTC, EEOC, SEC, OCC, FDA, CFPB) applying existing law to AI-related harms, a patchwork of state laws, and, since December 2025, an executive order directing agencies to challenge state AI laws administratively rather than new federal legislation replacing them.
What did President Trump's December 2025 AI executive order actually do?
Executive Order 14365, 'Ensuring a National Policy Framework for Artificial Intelligence,' signed 11 December 2025, set federal policy against a state-by-state AI law patchwork. It ordered the Attorney General to stand up an AI Litigation Task Force (created 9 January 2026) to challenge state AI laws in court, directed the Commerce Department to identify 'onerous' state laws within 90 days, and directed agencies to consider withholding federal broadband (BEAD) funding from states whose AI laws conflict with the administration's policy.
Has the executive order actually overturned any state AI laws?
Not as of mid-2026. The order works through litigation, funding conditions, and agency reinterpretation rather than automatic preemption. Legal trackers reported no state AI law had been successfully challenged or struck down under it by July 2026, though the AI Litigation Task Force was expected to file its first suit later in 2026, widely predicted to target Colorado. Until a court rules or a state repeals its own law, state AI statutes remain in force.
Is Congress fighting back against the executive order?
Yes. Senator Ed Markey introduced the States' Right to Regulate AI Act (S.3557) in the Senate just six days after the order, on 17 December 2025. Representative Don Beyer and colleagues followed with the House GUARDRAILS Act (H.R. 8031) on 20 March 2026, and Senator Brian Schatz introduced a Senate companion (S.4216) shortly after. Each of these bills would declare the executive order to have no force or effect and bar federal funds from being used to implement it. None had passed both chambers as of mid-2026.
How does the EU AI Act differ from the US approach?
The EU AI Act (Regulation (EU) 2024/1689) is a single, comprehensive, horizontal law that classifies every AI system by risk tier (prohibited, high-risk, limited-risk, minimal-risk) and applies extraterritorially to anyone placing AI on the EU market. The US has no equivalent statute; it relies on sector regulators applying existing law, plus, since December 2025, a federal push to limit state-level AI laws rather than a single national framework passed by Congress.
What happened to the US AI Safety Institute?
It was renamed the Center for AI Standards and Innovation (CAISI) in June 2025. Its mandate shifted from a safety-first framing toward supporting industry AI testing, security evaluation, and US influence over international AI standards, while remaining part of NIST within the Department of Commerce.
Last reviewed July 2026
This page is general information about What Is AI Regulation?, not legal, regulatory, or professional advice, and does not capture every nuance or exception. Requirements change and can be fact-specific. Always verify against primary sources and your own qualified legal counsel before relying on it.