On July 23, 2025, the president signed three AI-related Executive Orders (“E.O.s”) to accompany the recently released White House’s Artificial Intelligence (AI) Action Plan (“AI Action Plan”). These E.O.s seek to add clarity to, and drive forward, federal policy in the AI space.

While they all relate to AI, the E.O.s otherwise vary considerably in subject matter: “Accelerating Federal Permitting of Data Center Infrastructure”; “Promoting the Export of the American AI Technology Stack”; and “Preventing Woke AI in the Federal Government.”

As we noted in our July 24 blog, the White House is clearly determined to outpace other countries so that the U.S. benefits from any gains provided by AI through building AI infrastructure and bolstering AI-related exports. The Center for Data Innovation, from its perspective, stated in a press release that the actions pursued by the executive orders will advance U.S. goals of global AI dominance and enable the U.S. to better compete with China. 

More controversial is the E.O. on “woke AI,” which Congressional AI Caucus Chair Don Beyer (D-VA) and other Democratic members of the caucus called “counterproductive to responsible AI development and use, and potentially dangerous” in a July 24 press release. The Caucus went on to state: “To be clear, we support true AI neutrality—AI models trained on facts and science—but the administration’s fixation on ‘anti-woke inputs’ is definitionally not neutral.”

There will likely be significant ripple effects to the private sector that emanate from these E.O.s which we will continue to track. We discuss each of them in turn.

‘Accelerating Federal Permitting of Data Center Infrastructure’

The new E.O. on “Accelerating Federal Permitting of Data Center Infrastructure” revokes E.O. 14141 of January 14, 2025, “Advancing United States Leadership in Artificial Intelligence Infrastructure.” Pursuant to this new E.O., the administration aims to charts a course for “bold, large scale industrial plans”—including AI data centers and supporting infrastructure—to ensure U.S. leadership in manufacturing processes and technologies. The E.O. primarily seeks to realize AI data centers, through the establishment of “qualifying projects”; the easing of regulatory burdens; and the optimization of federally owned land and resources for the development of these centers. This policy push dovetails from the Project Stargate announcements we previously discussed here.

Qualifying Projects. Qualifying projects are defined as data center projects or covered component projects

• where the project sponsor has committed at least $500 million in capital expenditures;
• involving an incremental electric load of greater than 100 megawatts;
• protecting national security; or
• 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.”

Financial Support. The E.O. directs the Secretary of Commerce to launch an initiative to provide financial support for qualifying projects, which could include loans, grants, tax incentives, and offtake agreements, with “relevant agencies” tasked with identifying the potential financial support. However, as we discussed in our recent blog, the AI Action Plan’s admonition that federal funding should not be “directed toward states with burdensome AI regulations that waste these funds” suggest that federal financial support will be doled out to states with the least restrictive laws. 

Efficient Environmental Reviews. The White House directs certain federal agencies to take actions to expedite environmental reviews through identifying existing categorical exclusions or establishing new ones.

Fast Tracking: Qualifying projects may receive expedited permitting under the FAST-41 Framework.

Federal Lands Availability. Executive Order 14141 of January 14, 2025—according to a White House Fact Sheet—would have “saddled AI data center development on Federal lands with pages of DEI and climate requirements.” Among other things, the new E.O. promotes the use of Brownfield and Superfund sites for data center development. It also directs various agencies to authorize data center construction on “appropriate” federal lands and to authorize federal lands for the purpose.

‘Promoting the Export of the American AI Technology Stack’

The E.O. “Promoting the Export of the American AI Technology Stack” advocates decreasing international reliance on AI and an increasing global deployment of AI technologies originating from the United States. It calls for an American AI Exports Program, to support the development and deployment of U.S. full-stack AI export packages, within 90 days. The E.O. directs the Secretary of Commerce to solicit proposals from industry-led consortia, with requirements as outlined in the order, also to be submitted within 90 days (the Department of Commerce will consider proposals on a rolling basis).

Proposals selected by the Secretary of Commerce, in consultation with other agency heads, will be designated “priority AI Export Packages.” The E.O. further directs an Economic Diplomacy Action Group (EDAG) established in 2024 to mobilize federal financing tools in support of these packages. The Secretary of State, in consultation with the EDAG, shall

• develop and execute a federal government strategy to promote the export of American AI technologies and standards;
• align technical, financial, and diplomatic resources to accelerate deployment of priority AI export packages;
• coordinate U.S. participation in multilateral initiatives and country-specific partnerships for AI deployment and export promotion;
• support partner countries in fostering pro-innovation regulatory, data and infrastructure environments conducive to the deployment of American AI systems;
• analyze market access, including technological barriers to trade and regulatory measures that may impede the competitiveness of U.S. offerings; and
• facilitate investment in U.S. small businesses developing AI technologies and the manufacture of AI infrastructure, hardware, and systems.

Indeed, the White House Fact Sheet accompanying this order emphasizes “Making America the Global Leader in AI.”

‘Preventing Woke AI in the Federal Government’

The E.O. “Preventing Woke AI in the Federal Government” requires federal agencies to procure only those large language models (LLMs) that are consistent with “Unbiased AI Principles”—i.e., both truthful and ideologically neutral. According to the accompanying Fact Sheet:

• “Truth-seeking means that LLMs shall be truthful and prioritize historical accuracy, scientific inquiry, and objectivity, and acknowledge uncertainty where reliable information is incomplete or contradictory.”
• “Ideological neutrality means that LLMs shall be neutral, nonpartisan tools that do not manipulate responses in favor of ideological dogmas like DEI [diversity, equity, and inclusion] and that developers will not intentionally encode partisan or ideological judgments into an LLMs outputs unless those judgments are prompted by or readily accessible to the end user.”

The Office of Management and Budget (OMB), in consultation with others, must issue guidance to agencies regarding “Unbiased AI Principles” within 120 days.

In procuring federal contracts for LLMs, agency heads must include terms ensuring compliance with the “Unbiased AI Principles” and terms that hold vendors accountable for decommissioning costs if contracts are terminated due to noncompliance, following a reasonable period to cure. Existing contracts for LLMs will be revised—“to the extent practicable and consistent with contract terms”—to conform with this E.O. Finally, agencies must adopt procedures, within 90 days of the guidance noted above, to adopt procedures to ensure that LLMs procured by the agency comply with the “Unbiased AI Principles.”

Key Takeaways

This latest slate of E.O.s take one step further to advance the Trump Administration’s policies—promoting AI innovation, supporting the creation of AI infrastructure, and advocating for deregulation in these areas.

Although federal agencies undoubtedly will comply with these E.O.s, we anticipate potential legal challenges to these policies emanating from the legislative branch and the private sector. The E.O. on “Woke AI,” as seen by the response of the Democratic members of Congressional AI Caucus, will almost certainly be met with legal challenges. Some challenges are certain to arise as companies deploying AI navigate a delicate balance—between avoiding “Woke AI” and ensuring they do not violate federal and state anti-discrimination requirements.

Given the pronouncements of the E.O.s preceding the AI Action Plan and the concurrent publication of the trio of AI executive actions impacting society and the environment, it can be argued that the administration’s national AI policy calls for a comprehensive federal preemption of state regulatory actions by the de facto establishment of a federal standard of AI governance, without congressional mandate.

Thus, there is likely to be a 10th Amendment confrontation, with some states arguing that any federal standard may only represent the minimum of regulatory oversight. For example, federal minimum wage and overtime protection under the Fair Labor Standards Act (FLSA), as well as the Occupational Safety and Health Administration’s (OSHA’s) job safety protections, are laws that are eclipsed by those states that pursue more stringent protection for their workers than the federal government mandates. We can expect to see this effort with respect to implementation of state AI regulation, especially where state regulation calls for increased worker protections from job displacement or unlawful discrimination than envisioned by federal mandates for accelerated AI innovation over “diversity, equity and inclusion.”

Federal government contractors and subcontractors clearly within the purview of the AI Action Plan, according to the AI E.O.s as well as recent OMB guidance memos, will face added pressures to comply with procurement guidelines that will perforce impact their workforces. Indeed, compliance with “Unbiased AI Principles” contemplated in the E.O.s and requirements for “ideological neutrality” will create increased need for workforce training to avoid penalties of contract termination (“decommissioning”) and related noncompliance fees.

Moreover, the absence of any particular regulation is not the absence of risk. Common law and statutory tort exposure will be ever present, as well, for example, to the extent that AI input and/or output impacts data or personal privacy or is defamatory, hateful, or abusive.

We will continue to follow and blog on future developments.

Epstein Becker Green Staff Attorney Ann W. Parks contributed to the preparation of this post.