America's AI Action Plan is a Federal AI strategy issued by the White House in July 2025. It is organized around three pillars — accelerating AI innovation, building American AI infrastructure, and international AI diplomacy and security — and follows from Executive Order 14179 ("Removing Barriers to American Leadership in AI").
| Field | Detail |
|---|---|
| Issued | July 2025 |
| Authors | Michael Kratsios (OSTP), David Sacks (AI/Crypto Advisor), Marco Rubio (NSA) |
| Legal basis | Follows EO 14179 ("Removing Barriers to American Leadership in AI") |
Position in the US AI regulatory landscape
The Action Plan sits within a sequence of US executive actions on AI:
| Document | Date | Approach | |
|---|---|---|---|
| [[eo-13859 | EO 13859]] | Feb 2019 | Establish AI Initiative, pro-innovation principles |
| Biden EO 14110 | Oct 2023 | Safety/rights-focused regulation (rescinded) | |
| AI Action Plan | Jul 2025 | Three-pillar strategy | |
| [[eo-14365 | EO 14365]] | Dec 2025 | Federal preemption of state AI laws |
It reverses the Biden-era safety- and rights-focused posture of EO 14110 (rescinded), and is paired with the later federal-preemption order EO 14365 as a companion instrument of the same overall strategy.
Pillar I — Accelerate AI innovation
The first pillar directs deregulation: removing state and federal barriers to AI development, and conditioning federal funding on an AI-friendly state regulatory climate. It treats open-source and open-weight models as geostrategically important, calls for regulatory sandboxes and domain-specific standards, and frames a "worker-first AI agenda" built around retraining programs. It directs investment in interpretability, control, and robustness research through DARPA, and mandates federal employee access to frontier large language models.
The pillar also directs NIST to revise the AI RMF to remove references to DEI, misinformation, and climate.
Pillar II — Build American AI infrastructure
The second pillar addresses the physical underpinnings of AI deployment. It calls for streamlining NEPA permitting for data centers and energy projects, and for embracing nuclear fission, fusion, and enhanced geothermal generation. It continues the CHIPS program for domestic semiconductor manufacturing, directs construction of high-security data centers for military and intelligence-community use, and calls for training a workforce for AI infrastructure jobs such as electricians and HVAC technicians.
Pillar III — International AI diplomacy and security
The third pillar sets an export and diplomacy strategy: exporting the full AI technology stack to allies, countering Chinese influence in international AI-governance bodies, and strengthening export-control enforcement. It directs evaluation of frontier models for CBRNE and cyber risks, and requires nucleic acid synthesis screening as a biosecurity measure.
This pillar is the Trump administration's successor export-control strategy. It replaces the Biden three-tier country framework with a "full stack to allies plus counter-China" posture, following the rescission of the BIS AI Diffusion IFR on May 13, 2025, two days before that rule's May 15 compliance date.
Implementation
The Action Plan is a strategy document rather than a self-executing instrument; commentators counted more than 90 discrete policy recommendations across its three pillars, to be carried out through subsequent executive orders and agency action (Source: executivegov.com). It was released on July 23, 2025 alongside three same-day executive orders that began operationalizing it: EO 14320 ("Promoting the Export of the American AI Technology Stack," Pillar III), EO 14319 ("Preventing Woke AI in the Federal Government," a procurement-side companion to the Pillar I RMF direction), and an order accelerating federal permitting of data-center infrastructure (Pillar II) (Source: wiley.law).
Follow-on actions over the subsequent year extended each pillar. A November 2025 order launched the Department of Energy-led "Genesis Mission" to accelerate AI-driven scientific discovery (Pillar I/II) (Source: executivegov.com). In December 2025 the administration issued EO 14365, directing federal preemption of state AI laws — the harder-edged successor to the Action Plan's own funding-conditioning approach. In March 2026 the White House released a National Policy Framework for AI carrying legislative recommendations to Congress (Source: hklaw.com). In June 2026 a further order, on promoting advanced AI innovation and security, directed agencies to strengthen federal cybersecurity, expand AI-enabled defensive tools, and stand up a voluntary coordination framework for secure frontier-model deployment (EO — Promoting Advanced AI Innovation and Security (Trump, signed June 2, 2026)) (Source: mayerbrown.com).
WIRED reported on August 12, 2026 that White House officials are almost certain to revise the administration's AI guidelines and expand oversight of AI models, with open models among the additions under consideration, according to people familiar with the matter. No revised framework had been issued (Source: wired.com). Only the article's lede was retrievable, so which instrument is to be revised — the Action Plan itself, the March 2026 National Policy Framework, or agency guidance issued under either — is not established here, and neither is the form the expanded oversight would take.
Points of tension
The Action Plan states that AI will "complement, not replace" workers while simultaneously establishing rapid retraining programs aimed at displacement. Its direction to NIST to strip DEI, misinformation, and climate references from the AI RMF recasts the working definition of AI safety. Its conditioning of federal funding on states not maintaining "burdensome" AI regulations is a softer form of state preemption than the later EO 14365. Its biosecurity provisions align with recommendations from Dario Amodei, while its overall deregulatory posture conflicts with Amodei's graduated-response philosophy.
Several provisions carry directives whose implementation timelines were set out at issuance. The Action Plan text directs the NIST AI RMF revision, with a revised version anticipated before end-2026 (resolution date 2026-12-31). The federal-funding-conditioning provision was expected to draw a legal challenge from at least one (≥1) state AG within 12 months of the July 2025 issuance (by 2026-07). The Pillar III framing anticipates translation into two or more (≥2) concrete BIS export-control rule changes within 18 months (by 2027-01). Its workforce-retraining commitments under Pillar I were framed as reaching material funding through public Department of Labor, Treasury, or Commerce appropriations, on the order of more than $500M, before end-2027 (by 2027-12-31). Commentators tied to the procurement-versus-state-law tension have raised the prospect that the Action Plan, the Trump preemption EO, and the xAI v. Colorado litigation could together lead to at least one state AI law being preempted by a court ruling within 24 months (by 2027-12-31).
Relationships
- supersedes: BIS Framework for AI Diffusion — Interim Final Rule (Jan 13, 2025) — BIS rescinded the AI Diffusion IFR on May 13, 2025 (two days before its May 15 compliance date); the Action Plan's Pillar III (international AI diplomacy and security) is the Trump successor export-control strategy, replacing the Biden three-tier country framework with a "full stack to allies + counter-China" posture.
- related: Executive Order 14365 — Ensuring a National Policy Framework for AI — federal preemption EO; companion instrument of the same overall strategy.
- contradicts: Executive Order 14110 — Safe, Secure, and Trustworthy AI — reverses Biden-era safety/rights posture.