Removing Barriers to American Leadership in Artificial Intelligence: Inside the Executive Order Reshaping US AI Policy

Removing Barriers to American Leadership in Artificial Intelligence is Executive Order 14179, signed on 23 January 2025, which established a federal policy of sustaining and enhancing US global dominance in AI and revoked the previous administration's Executive Order 14110 on safe, secure, and trustworthy AI. The order's central claim is a policy judgement rather than a technical one: that ideologically motivated regulation and burdensome federal requirements were slowing American AI development, and that removing them would accelerate innovation, economic competitiveness, and national security. It directed the development of a national AI Action Plan and a review of every policy, rule, and guidance document issued under the revoked order. Whatever your position on it, this is the document that reset the baseline for AI compliance planning in the United States.

Quick Answer: Executive Order 14179, "Removing Barriers to American Leadership in Artificial Intelligence," was signed on 23 January 2025. It revoked Executive Order 14110, declared a policy of maintaining US global AI dominance free from ideological bias, and required an AI Action Plan plus a review of all prior AI-related federal guidance and regulations.

Building Compliance-Ready AI Products in a Shifting Policy Landscape

Deregulation at federal level does not reduce the engineering work of responsible AI deployment — it moves the obligation to states, sectors, and contracts. Companies still face state AI statutes, procurement requirements, and customer due-diligence questionnaires that ask how model output is logged, how personal data flows, and who can audit a decision. Digital agencies that build AI features end-to-end are increasingly writing those controls in from the first sprint. WebPeak, a worldwide full-service agency, approaches this through its AI engineering practice alongside product delivery on Next.js applications, so that audit logging, prompt versioning, and human-review checkpoints exist in the codebase rather than in a policy PDF. Their ongoing maintenance and support work matters just as much here, because AI compliance requirements change faster than most release cycles. You can review their full service range at webpeak.org.

What Executive Order 14179 Actually Changes

The order operates in three moves. First, it revokes: Executive Order 14110 of October 2023, which had imposed reporting requirements on developers of the most capable models and directed extensive agency rulemaking, was rescinded outright. Second, it instructs: agency heads, coordinated through the Assistant to the President for Science and Technology, the White House AI and Crypto Czar, and the National Security Advisor, were directed to identify and suspend, revise, or rescind actions taken under the revoked order that are inconsistent with the new policy. Third, it builds: it required an action plan to achieve the stated policy of American AI leadership, and directed revision of Office of Management and Budget guidance governing how federal agencies use and acquire AI.

The follow-through arrived in July 2025 as Winning the AI Race: America's AI Action Plan, which organised federal effort around accelerating innovation, building AI infrastructure including data centres and energy capacity, and leading on international AI diplomacy and export of the American AI stack. Updated OMB memoranda replaced the earlier federal-use guidance with a framing built around removing bureaucratic obstacles to agency AI adoption while retaining minimum practices for high-impact uses.

What the order does not do is create a comprehensive federal AI statute. An executive order cannot pre-empt state law by itself, and no general federal AI law exists. That gap is the single most consequential fact for businesses planning compliance.

What Businesses Should Do Now

The practical exposure for most companies is state and sectoral, not federal. These are the steps that hold up regardless of which way federal policy swings next.

• Map your AI systems by decision impact, not by technology. A model that screens job applicants or prices insurance carries regulatory weight; an internal summarisation tool usually does not. Most state frameworks and sector regulators draw the line at consequential decisions about people.
• Assume state law is the binding layer. Colorado's AI Act, Texas's AI governance statute, California's transparency requirements for frontier developers, and Illinois's and New York City's hiring-tool rules all operate independently of federal executive action.
• Keep an inventory with owners. Every AI system should have a named accountable person, a documented purpose, a data-source list, and a retention policy. This single artefact answers most customer security questionnaires.
• Log inputs, outputs, and human overrides. If you cannot reconstruct why a system produced a given result on a given date, you cannot defend it — to a regulator, an enterprise client, or a court.
• Do not dismantle existing governance because federal requirements loosened. Enterprise procurement and the EU AI Act still demand documentation, and rebuilding a retired governance process costs far more than maintaining it.
• Track infrastructure and energy policy if you train models. The AI Action Plan's data-centre and permitting emphasis directly affects compute availability and siting economics.

Comparing the Two Federal Approaches

Understanding what changed is easier as a side-by-side comparison of policy posture across the dimensions that affect real projects.

Dimension	Executive Order 14110 (2023, revoked)	Executive Order 14179 (2025, current)
Stated primary goal	Safe, secure, and trustworthy AI development	Sustained US global dominance in AI
Developer obligations	Reporting on the most capable models and safety test results	Reporting requirements rescinded
Agency direction	Extensive new rulemaking and standards work	Review, suspend, or rescind actions inconsistent with new policy
Required deliverable	Agency-by-agency implementation milestones	National AI Action Plan and revised OMB guidance
Treatment of state law	Not pre-empted	Not pre-empted by the order itself

Reading the Policy Signals: What Comes Next

Three verifiable developments define the current landscape. The AI Action Plan published in July 2025 committed the federal government to accelerating build-out of AI infrastructure and exporting the full American AI technology stack to allies — a shift from regulating models to competing on capacity. Separately, congressional attempts to impose a moratorium on state AI enforcement failed to survive the legislative process in 2025, leaving the patchwork intact. And the EU AI Act continued its phased application, with prohibited-practice rules effective in February 2025 and general-purpose AI obligations from August 2025, meaning any company serving European users faces obligations that US deregulation does not touch.

Beyond the documented record, one pattern from implementation work is worth stating plainly as analysis rather than dressing up as a statistic: companies that build governance as code — logging, evaluation harnesses, versioned prompts, override tracking — absorb regulatory change cheaply, while companies that build governance as documentation pay for it again with every jurisdiction and every enterprise deal. Federal deregulation lowers one cost and leaves that engineering cost exactly where it was.

The strategic read for most businesses is therefore unglamorous. Federal policy has become an accelerant for compute, energy, and procurement, not a source of relief from accountability. Teams scaling AI infrastructure under this environment should plan capacity deliberately, and guidance on cloud infrastructure planning is useful when workloads move from pilot to production scale.

Key Takeaways

• Executive Order 14179 was signed on 23 January 2025 and revoked Executive Order 14110, removing federal safety-reporting requirements for advanced model developers.
• It required a national AI Action Plan, delivered in July 2025 as "Winning the AI Race," focused on innovation, infrastructure, and international AI diplomacy.
• An executive order cannot pre-empt state law, so Colorado, Texas, California, and city-level AI rules remain the operative compliance layer for most companies.
• The EU AI Act continues to apply to companies serving European users regardless of US federal posture, with phased obligations from 2025 onward.
• Governance implemented in code — logging, evaluations, override tracking — survives policy reversals far better than governance held only in documentation.

Frequently Asked Questions

When was Removing Barriers to American Leadership in Artificial Intelligence signed?

The order was signed on 23 January 2025 and is designated Executive Order 14179. It took effect immediately, revoking Executive Order 14110 of October 2023 and directing federal agencies to review actions taken under the earlier order.

Does this executive order eliminate all AI regulation in the United States?

No. It removes federal requirements created by the prior executive order but does not repeal state AI laws, sector rules from agencies like the FTC or financial regulators, or existing anti-discrimination and consumer-protection law that already applies to automated decisions.

What is the AI Action Plan it required?

The plan, published in July 2025 as "Winning the AI Race: America's AI Action Plan," set federal priorities across accelerating innovation, building AI infrastructure including data centres and energy supply, and leading international AI diplomacy and technology export.

How does this affect companies operating in both the US and Europe?

Dual-market companies must meet the stricter standard. The EU AI Act imposes documentation, transparency, and risk-management duties on general-purpose and high-risk systems, so European obligations effectively set the internal baseline regardless of lighter US federal requirements.

Should we pause our AI governance programme because federal rules loosened?

No. Enterprise customers, insurers, and state regulators still require documented controls, and rebuilding a dismantled governance function is more expensive than maintaining it. Treat federal change as a shift in where obligations originate, not a removal of them.

Conclusion

If you take one decision from this order, make it this: build your AI compliance posture around state, sectoral, and contractual obligations, because those are the requirements that survived the federal reset and will outlast the next one. Executive Order 14179 changed the direction of federal AI policy decisively, but it left the layer that actually governs most business AI use untouched. Start by producing a single inventory of your AI systems with named owners, decision impact, data sources, and logging status. That document is the fastest route from policy uncertainty to defensible practice.