Last reviewed: April 26, 2026
Key Takeaways
- The White House released a National AI Legislative Framework on March 20, 2026 — a set of legislative recommendations sent to Congress, not an executive order or binding law.
- The framework’s most consequential proposal: federal preemption of state AI laws, which would override Colorado’s AI Act, Texas TRAIGA, and the growing patchwork of state regulation.
- Six stated principles cover children’s safety, community protection, intellectual property, free speech, innovation, and workforce development — but none are self-executing. Congress must legislate.
- The framework continues a 14-month deregulatory trajectory: Biden’s EO 14110 was revoked in January 2025, followed by innovation-focused EOs, the AI Action Plan, and an explicit preemption push starting in December 2025.
- For practitioners, the immediate question is not what this framework says, but what Congress does with it. State laws remain enforceable until preempted. Plan for both scenarios.
What the Framework Contains
On March 20, 2026, the White House sent Congress a document titled “National Policy Framework for Artificial Intelligence — Legislative Recommendations.” Two days earlier, on March 18, Senator Marsha Blackburn released the TRUMP AMERICA AI Act — a 291-page discussion draft that translates many of the framework’s principles into legislative text: duty of care for high-risk AI, mandatory training-data transparency, repeal of Section 230 for AI outputs, and a federal AI liability cause of action. It remains a discussion draft and has not been formally introduced. (White House)
The framework is organized around six principles:
1. Protect children and empower parents. Legislative recommendations for age-appropriate AI safeguards, parental controls, and guardrails against harmful AI interactions with minors.
2. Safeguard communities. Proposals addressing AI-enabled fraud, deepfakes, and critical infrastructure risks — building on the TAKE IT DOWN Act (signed May 2025) and FCC’s February 2024 ruling classifying AI-generated voices as illegal robocalls.
3. Respect intellectual property rights. Addresses AI training on copyrighted material and AI-generated content ownership — aligning with the Copyright Office’s three-part report series (2024-2025), which found that mere prompting is insufficient for copyright authorship.
4. Prevent censorship and protect free speech. Extends the “Preventing Woke AI” EO (14319, July 2025) into legislative territory. Proposes that AI systems should not “impose ideological bias” and that federal AI procurement should require “truth-seeking, ideologically neutral” models — language drawn from OMB M-26-04.
5. Enable innovation and ensure AI dominance. The framework’s centerpiece. Proposes regulatory sandboxes for AI developers, streamlined data center permitting (including on-site power generation), and — critically — federal preemption of state AI laws that the administration characterizes as burdensome.
6. Educate Americans and develop AI workforce. Workforce training and AI literacy initiatives.
What Is Binding vs. Aspirational
Here is the distinction that most reporting blurs: nothing in this framework is binding. It is a set of recommendations to Congress. No executive order was signed on March 20, 2026. No new regulations took effect. No deadlines were created.
What is binding from prior actions:
| Document | Date | Binding? | Status |
|---|---|---|---|
| EO 14179 (revoked Biden EO) | Jan 23, 2025 | Yes — revoked EO 14110 | Active |
| EO 14319 (“Preventing Woke AI”) | Jul 23, 2025 | Yes — directs agency AI procurement | Active |
| OMB M-25-21 (agency AI use) | Apr 3, 2025 | Yes — replaces M-24-10 | Active |
| OMB M-25-22 (AI procurement) | Apr 3, 2025 | Yes — Buy American for AI | Active |
| OMB M-26-04 (“Unbiased AI”) | Dec 11, 2025 | Yes — agencies must comply | Deadline passed (Mar 11, 2026) |
| EO 14365 (“National Policy Framework for AI”) | Dec 11, 2025 | Yes — preemption push + AG-led AI Litigation Task Force | Task Force operational; March 11 implementation deadlines missed |
| March 2026 Legislative Framework | Mar 20, 2026 | No — recommendations only | Awaiting Congress |
The practical implication: practitioners should track what Congress does with these recommendations, but cannot rely on them as current law. State AI laws (Colorado, Texas, Illinois, NYC, California, Washington) remain enforceable. EO 14365 (December 11, 2025) created the AG-led AI Litigation Task Force to challenge state laws, but that process operates through litigation, not automatic preemption. As of late April 2026, the order’s March 11 deadlines for the Commerce Department’s evaluation of state AI laws, the FTC’s preemption policy statement, and FCC rulemaking on a federal disclosure standard had all passed without deliverables (Silicon Report, April 2026).
What Changed from Prior Policy
The March 2026 framework is the latest step in a 14-month policy sequence. Here is the trajectory:
Phase 1 — Revocation (January 2025). EO 14179 revoked Biden’s EO 14110 on Day 3 of the Trump administration. The framing: Biden’s order imposed “ideological bias” and “unnecessary bureaucratic restrictions” on AI development. Agencies were directed to review and rescind all Biden-era AI actions.
Phase 2 — Replacement (April 2025). OMB issued M-25-21 and M-25-22, replacing Biden’s agency AI use and procurement guidance. The Brookings Institution noted that these memos “signal continuity in federal AI policy” despite the rhetorical shift — core governance concepts (risk assessment, monitoring, human oversight) survived, simplified into a single “high-impact AI” category replacing Biden’s multi-tiered system. (Brookings)
Phase 3 — Expansion (July 2025). A major policy day: the AI Action Plan (90 federal positions across three pillars), EO 14319 on “Preventing Woke AI,” data center permitting EO, and AI export stack EO. The administration established its pro-innovation, anti-regulation posture.
Phase 4 — Preemption push (December 2025). EO 14365, “Ensuring a National Policy Framework for Artificial Intelligence” (Dec 11, 2025), created the AI Litigation Task Force, conditioned federal broadband funding on compliance with national AI policy, and directed the AG to challenge state AI laws. This was the first explicit federal preemption signal. OMB M-26-04 implemented “Unbiased AI Principles” for federal procurement. The order set a March 11, 2026 deadline for Commerce, FTC, and FCC implementation actions; those deadlines passed in April 2026 with deliverables still unreleased.
Phase 5 — Legislative blueprint (March 2026). The framework translates executive-branch policy positions into Congressional recommendations. The preemption proposal would give these positions the force of statute — something executive orders alone cannot do.
One point that gets lost in the political framing: the practical governance requirements for federal agencies — risk assessment, human oversight, transparency, monitoring — largely survived the Biden-to-Trump transition. They were simplified and reframed, not eliminated. The real change is in the private-sector posture: from mandatory compliance obligations to voluntary frameworks, and from encouraging state regulation to actively opposing it.
Impact on Federal Procurement
If your organization sells AI to the US government, these are the current binding requirements:
OMB M-25-22 (April 2025) governs AI procurement. Solicitations issued on or after October 1, 2025 must comply. Key provisions:
- Buy American preference for AI systems
- Privacy protections for personally identifiable information
- Vendor performance monitoring
- Prohibition on using nonpublic agency data to train commercial AI without consent
OMB M-26-04 (December 2025) adds “Unbiased AI Principles.” Federal agencies must ensure AI contractors provide transparency, documentation, and disclosure. National security systems, open-source models, and models used incidentally for administrative purposes are exempt.
NIST AI RMF 1.0 remains the government’s voluntary risk management framework. While not legally binding, it is increasingly referenced in federal procurement requirements and state laws — see our NIST AI RMF practitioner’s guide. There is no “AI RMF 2.0” — NIST is pursuing incremental updates through companion profiles (Generative AI Profile, AI 600-1) and the Cybersecurity Framework Profile for AI (IR 8596, in draft).
Impact on Private Sector
The framework does not directly regulate the private sector. But three dynamics matter:
1. The preemption question. If Congress enacts federal preemption, state AI laws would be overridden — potentially eliminating Colorado’s impact assessment requirements, Texas’s prohibited-use framework, and California’s CPPA automated decision-making rules. Companies currently building compliance programs for state laws face a strategic question: invest now in state-by-state compliance, or wait for possible federal preemption?
Our assessment: build for state compliance. Federal preemption legislation faces significant political headwinds — the Senate killed a 10-year state AI moratorium 99-1 in July 2025. Congressional preemption would require bipartisan support that does not currently exist. The AI Litigation Task Force can challenge individual state laws through courts, but this is slow and uncertain.
2. The enforcement vacuum. The CFPB withdrew approximately 67 guidance documents in May 2025, including its key AI-specific circulars on algorithmic credit decisions and AI chatbots in financial services. The FTC’s posture has shifted toward reactive enforcement targeting demonstrable consumer harm. State attorneys general are filling this gap — Morgan Lewis characterized the dynamic as “AI Enforcement Accelerates as Federal Policy Stalls.” (Morgan Lewis, April 2026) For the broader private litigation picture, see our AI liability guide.
3. Nineteen new state laws. While the federal government debates preemption, states keep legislating. Through end of March 2026, 19 new state AI laws were enacted, including Washington’s chatbot safety law (HB 2225, with a private right of action), Oregon and Idaho chatbot bills, and Tennessee’s ban on AI posing as licensed mental health professionals. The 2026 state legislative sessions have already surpassed all of 2024 in AI bill volume: 1,561 bills across 45 states. (PluralPolicy)
US Federal Direction vs. EU AI Act
The contrast is structural, not just ideological. For the EU model in detail, see our EU AI Act Annex III guide.
| Dimension | US (March 2026 Framework) | EU (AI Act) |
|---|---|---|
| Legal form | Legislative recommendations (not law) | Regulation with direct effect (law since Aug 1, 2024) |
| Architecture | Voluntary + state patchwork + agency enforcement | Single horizontal law covering all sectors |
| Risk classification | No federal system; Colorado/Texas have state-level versions | Four tiers: prohibited, high-risk, limited, minimal |
| Enforcement | FTC, SEC, state AGs, private litigation — fragmented | National competent authorities + European AI Office |
| Compliance obligations | Vary by state and sector | Uniform across 27 member states |
| Federal preemption | Proposed — would override state laws | Built-in — EU regulation supersedes member state law |
| Timeline | Unknown — depends on Congress | Phased: prohibited (Feb 2025) → full (Aug 2027) |
The irony: the US is pursuing federal preemption precisely because its decentralized approach produced the fragmentation the administration now calls burdensome. The EU’s centralized approach avoids this problem by design — one law, one set of obligations, one timeline. For multinational companies operating in both jurisdictions, the EU AI Act remains the higher compliance bar regardless of what the US framework becomes.
What to Do Next
1. Track Congress, not the White House. The framework’s proposals only matter if Congress acts. Monitor Senator Blackburn’s TRUMP AMERICA AI Act discussion draft, the Senate Commerce Committee, and any preemption language attached to must-pass legislation (appropriations bills, NDAA).
2. Build for state compliance. Colorado’s AI Act takes effect June 30, 2026. Washington’s chatbot safety law is already in effect. New York’s RAISE Act compliance begins January 1, 2027. These deadlines are real. Federal preemption is hypothetical.
3. Align your AI governance with NIST AI RMF. Whether or not the framework becomes law, the NIST AI Risk Management Framework is increasingly referenced in federal procurement, state legislation, and international standards. Implementing its four core functions (Govern, Map, Measure, Manage) positions your organization for any regulatory scenario.
Related Reading
US AI Regulation Series:
- NIST AI Risk Management Framework — The federal procurement governance baseline
- AI Liability in the US — Where the Blackburn TRUMP AMERICA AI Act would land
- Texas TRAIGA Compliance Guide — A state law potentially preempted
- Colorado AI Act 2026: What Developers and Deployers Must Do — the closest US state analogue to EU Annex III; KILO draft watch
- SEC and AI: What Financial Firms Need to Know — sector enforcement playbook, six AI washing cases
- EU vs US AI Regulation: The Definitive Comparison — cross-jurisdiction picture, NIST RMF as bridge, Digital Omnibus status
- Colorado High-Risk AI Classifier (interactive tool) — test your AI system against SB 24-205 in six questions
- Illinois AI Employment Law 2026: AIVICA + HB 3773 — disparate-impact standard with private right of action via IHRA
- NYC Local Law 144: AI Bias Audit Guide — first US mandatory bias audit; DCWP enforcement now active
- FDA AI Medical Devices: PCCP + EU AI Act Comparison — 1,451+ AI devices authorized; PCCP framework for adaptive AI
EU Comparison:
- EU AI Act Annex III — The risk-classification model the US framework rejects
Sources
Primary Sources
- White House, “National Policy Framework for Artificial Intelligence — Legislative Recommendations,” March 20, 2026 (whitehouse.gov)
- EO 14179, “Removing Barriers to American Leadership in Artificial Intelligence,” January 23, 2025 (Federal Register)
- EO 14319, “Preventing Woke AI in the Federal Government,” July 23, 2025
- OMB M-25-21, “Accelerating Federal Use of AI,” April 3, 2025
- OMB M-26-04, “Unbiased AI Principles,” December 11, 2025
Analysis
- Brookings, “New OMB memos signal continuity in federal AI policy” (brookings.edu)
- Sidley, “Unpacking the December 11, 2025 Executive Order” (sidley.com)
- Morgan Lewis, “AI Enforcement Accelerates as Federal Policy Stalls,” April 2026 (morganlewis.com)
- PluralPolicy, “AI Governance Watch: April 2026” (pluralpolicy.com)
- Silicon Report, “Trump administration AI deadlines slip,” April 2026 (siliconreport.com)
This article provides general information about AI regulation and does not constitute legal advice. Laws and policies change frequently. Consult qualified legal counsel for compliance decisions specific to your organization.