President Trump Signs Three Executive Orders Relating to Artificial Intelligence

August 08, 2025

By Amir R. Ghavi,Quinn Dang,Katie Katsuki and Howard Glucksman

On July 23, 2025, just hours after releasing the AI Action Plan, President Donald Trump signed three executive orders aimed at accelerating data center construction, expanding the global reach of U.S. AI technology and addressing allegedly biased or “woke”[1] AI models in federal procurement. Announced during an AI summit in Washington, D.C., these orders align with the AI Action Plan’s goal of reducing regulatory barriers, ramping up AI infrastructure and promoting U.S.-made technology abroad.

Accelerating Federal Permitting of Data Center Infrastructure[2]

Executive Order 14318, “Accelerating Federal Permitting of Data Center Infrastructure,” aims to accelerate the development of AI-related infrastructure in the United States by reducing regulatory barriers and deploying U.S. government resources. The executive order targets the buildout of AI data centers and related materials and critical infrastructure that are required to build out data centers such as energy infrastructure, including high-voltage transmission lines; natural gas, coal, nuclear and geothermal equipment; semiconductor and semiconductor materials; networking equipment (switches and routers); and data storage systems, including integrated services that work with cloud service providers. To accomplish this, the Trump administration plans to use federal land and streamline permitting to fast-track construction efforts and sets forth the following mandates with respect to “Qualifying Projects”:

Financial Support
- The secretary of commerce, in consultation with the Office of Science and Technology Policy (OSTP) and other relevant agencies, is tasked with launching an initiative to provide financial support for Qualifying Projects, which could include loans, loan guarantees, grants, tax incentives and offtake agreements to such eligible projects.[3] Qualifying Projects include data centers (or projects involving infrastructure required to build out such data centers) that have a commitment of least $500 million in capital, an incremental electric load addition of greater than 100 MW or protect national security.

Efficient Environmental Reviews
- By August 2, 2025, each of the relevant agencies must report existing National Environmental Policy Act (NEPA) categorical exclusions[4] to the Council on Environmental Quality (CEQ) to streamline environmental reviews. CEQ is directed to work with federal agencies to create new categorical exclusions specifically for actions tied to Qualifying Projects that typically do not have a significant effect on the human environment. Additionally, the order avoids an automatic environmental review of all funded projects by establishing a 50% federal funding threshold to determine a “major Federal action” under NEPA.[5]
- The Environmental Protection Agency (EPA) must review regulations under existing environmental laws to facilitate efficient permitting.[6] By January 19, 2026, the EPA must also identify Brownfield[7] and Superfund Sites suitable for reuse in Qualifying Projects and issue guidance on expediting their redevelopment.[8]
Efficiency and Transparency Through FAST‑41
- Within 30 days of an agency identifying a Qualifying Project, the executive director of the Federal Permitting Improvement Steering Council (FPISC) may designate it as a transparency project and, within 30 days of receiving such agency notification, publish it on the Permitting Dashboard established under Section 41003(b) of FAST-41, including schedules for expedited review. In consultation with Project Sponsors, the executive director is also directed to expedite the transition of eligible projects to FAST-41 covered project[9] status. For projects that do not meet the covered project status, FPISC may explore alternative paths to designation under 42 U.S.C. 4370m(6)(A)(iv) to ensure inclusion and further streamline review.

Federal Lands Availability
- The Department of the Interior and the Department of Energy are directed to identify and authorize appropriate federal sites for development. These authorizations must follow applicable laws and involve consultation with industry and the Department of Commerce. Once sites are identified by the secretaries of the interior and energy, the designated action agency must initiate a consultation with the secretary of the interior, secretary of commerce, or both, addressing common construction activities for Qualifying Projects expected over the next 10 years.[10]
- The secretary of defense is also directed to identify suitable locations on military installations for infrastructure supporting Covered Components. These lands shall be made available with competitive leases for use in Qualifying Projects, provided that they align with the Department of Defense’s energy, workforce and mission needs, and do not compromise security or force protection.

Promoting the Export of U.S. AI Technology Stack

Executive Order 14320, “Promoting the Export of the American AI Technology Stack,” underscores the strategic importance of AI in shaping the future of global power dynamics, economic growth and national security. The executive order recognizes AI as “a foundational technology that will define the future of economic growth, national security, and global competitiveness for decades to come.” It sets forth its strategy to “preserve and extend American leadership in AI and decrease international dependence on AI technologies developed by [U.S.] adversaries by supporting the global deployment of U.S.-origin AI technologies.”[11]

To accomplish this, the executive order establishes a coordinated national effort through the American AI Exports Program (Exports Program). By October 21, 2025, the secretary of commerce, in consultation with the secretary of state and the director of the OSTP, is directed to establish and implement the Export Program to support the development and deployment of United States full-stack AI export packages. The secretary of commerce will publicly solicit proposals from industry-led consortia seeking to participate in the program.

Each proposal must include a comprehensive, full-stack AI technology package, which covers:[12]
- AI-optimized computer hardware and infrastructure (e.g., chips, servers and accelerators), data center storage, cloud services and networking, as well as identifying whether and to what extent such items are manufactured in the United States
- Data pipelines and labeling systems
- AI models and systems
- Cybersecurity and physical security measures for AI models and systems
- AI applications for specific use cases (e.g., software engineering, education, healthcare, agriculture or transportation)
- Specific target countries or regional blocs for export engagement
- Business and operational models to explain, at a high level, which entities will build, own and operate data centers and associated infrastructure
- Requested federal incentives or support
- Compliance with all relevant U.S. export controls regimes, outbound investment regulations and end user policies, including under the Export Control Reform Act and “relevant guidance” from the Bureau of Industry and Security within the Department of Commerce

The secretary of commerce shall, in consultation with the secretaries of state, defense and energy and the director of the OSTP, evaluate the submitted proposals for inclusion under the program. The proposals selected shall be designated as “priority AI export packages” and supported with financial tools specified under Section 4 of the executive order, which describes “Federal Financing Tools” to be mobilized under the authority of the “Economic Diplomacy Action Group” (EDAG) established under presidential memorandum of June 21, 2024. The EDAG, chaired by the secretary of state, will lead this effort by aligning technical, financial and diplomatic resources. The secretary of state is also tasked with promoting U.S. AI standards abroad and helping partner countries establish favorable environments for deploying U.S. AI systems. Lastly, the executive order sets forth how the U.S. government will mobilize federal financing tools, such as loans, equity investments and technical assistance, to support selected AI export packages.

Preventing Potentially Biased AI in the Federal Government

Executive Order 14319, “Preventing Woke AI in the Federal Government,” aims to address concerns by the Trump administration that artificial intelligence systems, particularly LLMs, may be compromised by ideological bias, which is described in the executive order as principles associated with “diversity, equity, and inclusion” (DEI). According to the executive order, when ideological biases or social agendas are built into AI models, they can distort the quality and accuracy of the output. These elements, the executive order states, replace “a commitment to truth in favor of preferred outcomes” and thus pose “an existential threat to reliable AI.”[13]

To address this, the executive order establishes two core “Unbiased AI Principles” for federal procurement of AI systems:

Truth-Seeking: LLMs must prioritize “historical accuracy, scientific inquiry, and objectivity” and acknowledge uncertainty where applicable.
Ideological Neutrality: LLMs must function as “neutral, nonpartisan tools” that do not “manipulate responses in favor of ideological dogmas such as DEI.”[14]

The executive order directs the Office of Management and Budget (OMB) to issue guidance within 120 days to federal agencies on implementing the Unbiased AI Principles, including specific factors that each federal agency must consider in determining whether potential vendors comply with the Unbiased AI Principles. The executive order mandates that such guidance must (1) account for technical limitations in complying with the order; (2) avoid requiring disclosure of specific model weights or other sensitive technical data; (3) afford latitude for vendors; and (4) make exceptions for national security systems.[15] Following the release of such guidance, federal agencies must contractually require providers of LLMs to comply with the Unbiased AI Principles and amend existing agreements to require the same.[16]

As with all federal procurement standards, we expect this executive order to significantly impact the AI market as the federal government’s significant purchasing power allows it to shape norms.[17]

Conclusion

In a departure from the previous administration’s AI posture, the Trump administration has pivoted U.S. AI policy toward what it considers to be strategic advantage and opportunity. Neither the AI Action Plan nor the executive orders attempt to define “frontier models” (those trained at the highest levels of computational power), which is also a notable omission by the previous administration. The absence of any meaningful regulation on such models reinforces the sense that the U.S. is more focused on accelerating its leadership in the global AI race than limiting AI development out of fear of their potential capabilities. While the previous administration has made public commitments regarding responsible use of AI, particularly around mitigating bias and discrimination, the Trump administration’s evolving policy emphasis on content neutrality and procurement-driven incentives suggests a shift away from such guardrails to broader strategic objectives, especially those that align with global competitiveness and national interest.

Additionally, the “Promoting the Export of American AI Technology Stack” executive order is the first formal action the Trump administration has taken to repudiate the Biden administration’s AI diffusion policy, as well as its restrictions on the development of AI. Coupled with the administration’s decision to permit the export of NVIDIA’s H20s to China, a decision that will further advance inferencing capabilities on U.S. chips, this executive order is a clear signal that the U.S. is willing to take an “offensive” approach to ensuring its lead in AI, particularly by helping U.S. companies sell abroad and edge out competition from adversarial countries. Although the order references the term “export,” it remains silent on how the Trump administration will streamline approvals for AI chip exports to countries that must still navigate the complex licensing requirements for advanced computing items.

While some aspects of the Trump administration’s long-term export strategy remain unclear, the overall direction is clear: the Trump administration is taking a competitive stance in the global AI arena and betting on the U.S. to lead.

[1] Although not specifically defined in the executive order, “woke” refers to the embedding of ideological biases or social agendas, particularly those aligned with diversity, equity and inclusion, into AI models in ways that compromise factual accuracy, objectivity and truth-seeking.

[2] This executive order revokes Executive Order 14141 of January 14, 2025 (“Advancing United States Leadership in Artificial Intelligence Infrastructure”).

[3] A “Qualifying Project” means: (i) a Data Center Project or Covered Component Project for which the Project Sponsor has committed at least $500 million in capital expenditures as determined by the secretary of commerce; (ii) a Data Center Project or Covered Component Project involving an incremental electric load addition of greater than 100 megawatts (MW); (iii) a Data Center Project or Covered Component Project that protects national security; or (iv) a Data Center Project or Covered Component Project that has otherwise been designated by the secretary of defense, the secretary of the interior, the secretary of commerce or the secretary of energy as a “Qualifying Project.” “Data Center Project” means a facility that requires greater than 100 MW of new load dedicated to AI inference, training, simulation or synthetic data generation. “Project Sponsor” means the lead sponsor providing financial and other support for a Data Center Project or Covered Component Project, as determined by the secretary of defense, the secretary of the interior, the secretary of commerce or the secretary of energy, as appropriate. “Covered Component” means the materials, products and infrastructure that are required to build Data Center Projects or otherwise upon which Data Center Projects depend on.

[4] Categorical exclusions refer to a NEPA mechanism that exempts certain projects from detailed environmental review.

[5] This means such federal funding will not automatically trigger a full environmental review, establishing a presumption that if federal funds account for less than 50% of a project’s total cost, then there is no substantial federal control.

[6] These laws include Clean Air Act (42 U.S.C. 7401 et seq.); the Clean Water Act (33 U.S.C. 1251 et seq.); the Comprehensive Environmental Response, Compensation and Liability Act (42 U.S.C. 9601 et seq.); and the Toxic Substances Control Act (15 U.S.C. 2601 et seq.).

[7] As defined in 42 U.S.C., “Brownfield sites” means real property, the expansion, redevelopment or reuse of which may be complicated by the presence or potential presence of a hazardous substance, pollutant or contaminant (§ 9601(39)). A “Superfund Site” means any site where action is being taken pursuant to 42 U.S.C. 9604, 9606 or 9620 (referring to federal authorities under the Comprehensive Environmental Response, Compensation and Liability Act (CERCLA) to conduct or compel hazardous site cleanups, including emergency responses (§ 9604), enforcement actions against responsible parties (§ 9606) and cleanup obligations at federal facilities (§ 9620)).

[8] Additionally, by January 19, 2026, the Army Corps of Engineers is directed to assess whether new or revised nationwide permits are needed to facilitate development under Section 404 of the Clean Water Act.

[9] The term “covered project” in the context of FAST-41 means any activity in the United States that requires authorization or environmental review by a federal agency involving construction of infrastructure for renewable or conventional energy production, electricity transmission, surface transportation, aviation, ports and waterways, water resource projects, broadband, pipelines, manufacturing, semiconductors, artificial intelligence and machine learning, high-performance computing and advanced computer hardware and software, quantum information science and technology, data storage and data management, cybersecurity, carbon capture, energy storage or any other sector as determined by a majority vote of the FPISC that (i)(I) is subject to NEPA, (II) is likely to require a total investment of more than $200,000,000 and (III) does not qualify for abbreviated authorization or environmental review processes under any applicable law; (ii) is covered by a programmatic plan or environmental review developed for the primary purpose of facilitating development of carbon dioxide pipelines; (iii) is (I) subject to NEPA, (II) sponsored by an Indian Tribe (as defined in Section 5304 of Title 25), an Alaska Native Corporation, a Native Hawaiian organization (as defined in Section 7517 of Title 20), the Department of Hawaiian Home Lands or the Office of Hawaiian Affairs and (III) located on land owned or under the jurisdiction of the entity that sponsors the activity under subclause (II); or (iv) is subject to NEPA and the size and complexity of which, in the opinion of the FPISC, make the project likely to benefit from enhanced oversight and coordination, including a project likely to require (I) authorization from or environmental review involving more than two federal agencies, or (II) the preparation of an environmental impact statement under NEPA.

[10] As identified through the process set forth in the Endangered Species Act, 16 U.S.C. 1531–1544.

[11] The policy goals cited in the executive order is aligned with several policy actions set forth in the AI Action Plan.

[12] Proposals are due within 90 days of the public RFP and will be reviewed on a rolling basis. A final evaluation will be conducted by the secretary of commerce in coordination with the secretaries of state, defense and energy and the director of the OSTP.

[13] The executive order cites examples of what it considers to be “ideological bias.” In one case, an AI system altered the race or gender of historical figures such as the pope, the Founding Fathers and Vikings when generating images, allegedly prioritizing DEI guidelines over historical accuracy. Another model declined to create images celebrating white individuals’ achievements, though it fulfilled similar requests for other racial groups. A third example involved an AI refusing to “misgender” someone, even in a hypothetical scenario involving a nuclear apocalypse. According to the Trump administration, federal agencies have a responsibility to ensure the tools they procure are accurate, neutral and free from ideological distortion.

[14] The executive order describes ideological neutrality as the absence of “intentionally encoded partisan or ideological judgments,” unless explicitly prompted by the user.

[15] The term “national security system” means any information system (including any telecommunications system) used or operated by an agency or by a contractor of an agency, or other organization on behalf of an agency, (i) the function, operation or use of which (I) involves intelligence activities, (II) involves cryptologic activities related to national security, (III) involves command and control of military forces, (IV) involves equipment that is an integral part of a weapon or weapons system, or (V) subject to subparagraph (B), is critical to the direct fulfillment of military or intelligence missions; or (ii) is protected at all times by procedures established for information that have been specifically authorized under criteria established by an executive order or an act of Congress to be kept classified in the interest of national defense or foreign policy. Subparagraph (A)(i)(V) does not include a system that is to be used for routine administrative and business applications (including payroll, finance, logistics and personnel management applications).

[16] Certain states have also considered enacting similar provisions. For example, a March 2025 draft of the Texas Responsible AI Governance Act (TRAIGA) contained a restriction on the use of AI to “block, ban, remove, de-platform, demonetize, debank, de-boost, restrict, or otherwise discriminate against a used based on the user’s political speech,” which was later removed.

[17] Earlier this month, the Department of Defense announced partnerships with four frontier model-providers worth as much as $200 million each.

Click here for a PDF of the full text