On July 23, 2025, President Trump announced a sweeping new AI strategy, releasing the highly anticipated “Winning the AI Race: America’s AI Action Plan” and signing three executive orders regarding federal procurement, international exports and infrastructure development for AI. This coordinated policy rollout directly affects how AI is built, deployed, and regulated in the United States and abroad. Together, the new policies aim to accelerate U.S. AI innovation and reduce regulatory barriers.

In this alert we unpack what these developments mean for businesses navigating today’s fast-moving AI policy environment.

America’s AI Action Plan

Key Elements of the AI Action Plan

The White House released its AI Action Plan in July 2025 to position the U.S. as the global leader in artificial intelligence. The Plan is guided by three core pillars:

Accelerating Innovation: Removing regulatory barriers to fast-track AI development, promote open-source models, build high-quality scientific datasets, and support rapid federal adoption of AI, especially in areas like defense.
Building American AI Infrastructure: Modernizing and expediting permits for data centers and semiconductor manufacturing facilities and investing in a skilled workforce to support core AI needs such as electricians and HVAC technicians.
Leading AI Diplomacy and Security Globally: Expanding U.S. AI exports, safeguarding American technology from misuse and theft, and coordinating national security policies around AI advancements.

What Should the Private Sector Know About the AI Action Plan?

The AI Action Plan represents a comprehensive federal strategy to accelerate AI innovation, reduce regulatory barriers, and assert U.S. leadership in the global AI landscape. For the private sector, this means a likely reduction in regulatory burdens as well as expanded opportunities for research, funding, and workforce support. The following summarizes what businesses should be aware of:

Deregulation and Regulatory Streamlining: The Action Plan emphasizes the removal of “red tape” and onerous regulations that could stifle AI innovation. The Office of Science and Technology Policy (OSTP) is tasked with soliciting input from businesses and the public to identify federal regulations that hinder AI development and adoption.
Federal Pressure on State-Level AI Regulation: Federal funding may be tied to a state’s AI regulatory climate, potentially limiting support for states with restrictive AI laws. The Office of Management and Budget (OMB) will coordinate with federal agencies to consider a state’s AI regulatory climate when allocating funding, potentially limiting support to states with regimes deemed overly restrictive. Additionally, the Federal Communications Commission (FCC) will assess whether state AI regulations interfere with its federal mandates, signaling possible preemption or challenges to state-level rules that conflict with federal priorities.
Review and Potential Rollback of FTC Actions: The Action Plan calls for a comprehensive review of all Federal Trade Commission (FTC) investigations, orders, and consent decrees from the previous administration that may impose unduly burdens on AI innovation. Where necessary, the government will seek to modify or set aside orders that are seen as inhibiting technological progress.
Emphasis on Preventing “Woke” AI: The government intends to revise the National Institute of Standards and Technology (NIST) AI Risk Management Framework to remove references to misinformation, diversity, equity, and inclusion (DEI), and climate change. Federal procurement guidelines will be updated to ensure that only AI systems deemed objective and free from “top-down ideological bias” are eligible for government contracts. There is also a focus on evaluating Chinese AI models for alignment with state censorship.
Support for Open-Source and Open-Weight AI: The Action Plan identifies open-source and open-weight AI models as important tools for fostering innovation, particularly for startups, academic researchers, and businesses that handle sensitive data. The government views these models, which are freely available for anyone to download and modify, as assurances that small and medium-sized businesses, as well as the research community, are not limited to a narrow set of proprietary options from closed model providers. The endorsement of open-source is viewed as essential for fostering a dynamic, competitive AI ecosystem and positioning American open models as potential global standards.
Support for Research and Workforce Development: New initiatives will expand access to large-scale computing resources for startups and academic researchers, addressing a key barrier to entry in the AI sector. The government will also introduce regulatory sandboxes and sector-specific initiatives to accelerate AI adoption, particularly in traditionally cautious industries like healthcare. Furthermore, there is a strong emphasis on workforce development, with initiatives aimed at increasing AI literacy, retraining workers, and integrating AI education into broader workforce programs.

Executive Order: Preventing Woke AI in the Federal Government

Key Elements of the Order

The “Preventing Woke AI in the Federal Government” executive order establishes new requirements for artificial intelligence (AI) systems, specifically large language models (LLMs), procured by federal agencies. This is the first time the U.S. government has explicitly sought to influence the ideological behavior of AI.

The order’s primary goal is to ensure that AI models used by the federal government are free from what it describes as “ideological biases or social agendas,” particularly those linked to diversity, equity, and inclusion (DEI) initiatives. Under the order, federal agencies may only procure LLMs that comply with two “Unbiased AI Principles”:

Truth-seeking: AI must prioritize historical accuracy, scientific inquiry and objectivity, and must acknowledge uncertainty when information is incomplete or contradictory.
Ideological Neutrality: AI must remain neutral and nonpartisan and may not intentionally encode partisan or ideological judgments into outputs unless specifically prompted by the user.

What Should Tech Companies Seeking to Contract with the Federal Government Know About the Order?

Tech companies seeking to contract with the federal government for LLMs will face several new compliance obligations under this executive order, including:

Contractual Requirements: All new federal contracts for LLMs must include terms requiring compliance with the Unbiased AI Principles. This means vendors must ensure their models are both “truth-seeking” and “ideologically neutral” as defined by the order. If a vendor’s model is found noncompliant and fails to cure the issue within a reasonable period, the agency may terminate the contract and charge decommissioning costs to the vendor.
Transparency and Documentation: Vendors may be required to provide documentation demonstrating how their models meet the Unbiased AI Principles. This could include disclosure of system prompts, model specifications, and evaluation results. However, the order seeks to avoid requiring disclosure of sensitive technical data such as model weights, where practicable.
Ongoing Oversight: Agencies are expected to adopt procedures to ensure ongoing compliance, which may include regular audits or assessments. Existing contracts may also be revised to incorporate these requirements where feasible.

Additionally, the executive order will likely have significant implications for how AI models are designed, trained, and maintained for federal use:

Model Design and Training: Developers will need to carefully review their training data, system prompts, and alignment strategies to ensure outputs do not reflect or prioritize “DEI-related” concepts or other “ideological dogmas” as defined by the order. This may require re-engineering models to avoid certain types of content moderation or bias mitigation techniques that have become standard in the industry, especially those aimed at addressing historical or social inequities. Vendors may need to maintain separate versions of their models—one tailored to federal requirements and another for commercial or public use—if the federal standards diverge significantly from industry norms or international best practices.
Risk of Ambiguity and Subjectivity: The definitions of “ideological neutrality” and “truth-seeking” are open to interpretation and may evolve with future OMB guidance or changes in administration priorities. This creates legal and operational uncertainty for vendors, who may face challenges in determining what constitutes prohibited “ideological bias” versus permissible factual or scientific content.
Privacy and Data Handling: While the order does not directly impose new privacy requirements, the increased documentation and transparency obligations may require vendors to share more information about their models’ inner workings. Vendors should ensure that any disclosures do not inadvertently expose proprietary information or user data and should negotiate contract terms to protect trade secrets and intellectual property.

Executive Order: Promoting the Export of the American AI Technology Stack

Key Elements of the Order

The “Promoting the Export of the American AI Technology Stack” Executive Order aims to strengthen the United States’ global leadership in AI by supporting the export of “full-stack”—encompassing hardware, data infrastructure, AI model, cybersecurity measures, and sector-specific applications—American AI technologies to allies and partners around the world.

What Are the Requirements for Industry-Led Consortia Seeking to Export AI Technology Packages?

The executive order establishes the American AI Exports Program, led by the U.S. Department of Commerce. The Program encourages industry-led consortia to develop and export “full-stack” AI technology packages. Proposals for such packages must:

Include a combination of AI-optimized hardware (chips, servers, accelerators), data pipelines and labeling systems, cloud computing infrastructure, AI models and systems, cybersecurity measures, and AI applications for sectors such as software engineering, healthcare, education, agriculture, and transportation.
Identify specific countries or regions for export, supporting U.S. interests abroad.
Comply with U.S. export controls, investment restrictions, and end-user requirements.

What Support Does the U.S. Government Offer for These Export Packages?

The Secretary of Commerce will select proposals in consultation with the Secretary of State, the Secretary of Defense, the Secretary of Energy, and the Director of OSTP. These “priority AI export packages” gain access to a range of federal resources, including direct loans, loan guarantees, equity investments, and technical assistance, through the Economic Diplomacy Action Group (EDAG). The initiative also emphasizes support for small businesses and fosters pro-innovation regulatory environments in partner countries.

Executive Order: Accelerating Federal Permitting of Data Center Infrastructure

Key Elements of the Order

The “Accelerating Federal Permitting of Data Center Infrastructure” Executive Order is designed to fast-track the development of large-scale AI data centers, high-voltage transmission lines, and other essential infrastructure and equipment. This order aims to boost U.S. technological dominance, economic growth, and national security by streamlining federal permitting and leveraging federal lands for advanced data center projects.

What Types of Data Center Projects are Eligible for Accelerated Permitting and Federal Support Under the New Executive Order?

A “qualifying project” under this order includes:

Large-scale data centers with over 100 megawatts (MW) of new electric load dedicated to AI training, inference, simulation, or synthetic data generation.
Facilities manufacturing key components for data centers, such as semiconductors, energy infrastructure (transmission lines, substations, turbines), networking equipment, and data storage systems.
Projects with at least $500 million in capital investment, or those designated as critical to national security by relevant federal agencies.
Projects that add significant new electric load or are specifically designated by the Secretaries of Commerce, Defense, Energy, or the Interior.

What Specific Hurdles Does the New Executive Order Aim to Eliminate?

Lengthy and complex NEPA reviews: The order instructs federal agencies to expand their use of categorical exclusions under laws such as the National Environmental Policy Act (NEPA). The order also clarifies that federal financial assistance (such as grants or loans) that covers less than 50% of total project costs will not trigger NEPA’s “major Federal action” requirement. This reduces or eliminates the need for lengthy and complex environmental impact statements for qualifying data center projects.
Slow permitting under environmental laws and review: The Environmental Protection Agency (EPA) is directed to speed up permitting processes under the Clean Air Act, Clean Water Act, and other relevant environmental laws. Furthermore, the order grants qualifying data center projects the status of “covered projects” under the FAST-41 framework. This ensures accelerated, coordinated federal permitting for projects exceeding $500 million in cost or using more than 100 megawatts of electricity.
Delays in reusing contaminated or underutilized sites: The EPA is also tasked with identifying suitable Brownfield and Superfund sites for Qualifying Projects. This aims to make it easier and faster for developers to repurpose these sites, reducing pressure on undeveloped land and speeding up project timelines.
Delays in biological and water permitting: The order mandates programmatic consultation under the Endangered Species Act for common construction activities, allowing for more efficient, large-scale review rather than project-by-project consultation.
Limited access to federal and military lands: The Department of the Interior and Department of Energy are instructed to proactively identify and offer appropriate authorizations for sites on federal lands, in consultation with industry and other agencies.

How Does This Differ from the Previous U.S. AI Infrastructure Executive Order That Was Repealed?

The July 2025 order repeals an earlier executive order from January 2025, “Advancing Federal Permitting of Data Center Infrastructure.” Unlike the previous order, which required environmental reviews as well as alignment with clean energy goals, the new order eliminates or reduces many of these regulatory requirements. This represents a shift from a guardrail-focused approach to a more streamlined process aimed at accelerating data center development and minimizing administrative barriers.

Sunwoo Lee (Summer Associate) contributed to this article.

[View source.]