On July 23, the White House unveiled the administration’s awaited AI plan, “Winning the [AI] Race: America’s AI Action Plan” (AI Plan). Although the AI Plan features more than 90 policy actions suggested by the administration, the AI Plan embraces policies consistent with the administration’s prior activities on AI and revisits consistent themes driving the direction of US AI policy.
What’s In the AI Plan
The AI Plan itself contains three pillars of activity to drive US AI policy: (1) accelerate AI innovation, (2) build AI infrastructure, and (3) lead international AI diplomacy and security. However, across these three pillars there are four consistent overarching themes: regulation, investment, research and standardization, and cybersecurity.
Regulation and Deregulation. Regulation is woven throughout all three pillars of the AI Plan and indeed, deregulation is among the first policy initiatives of the AI Plan. This deregulation comes in several different forms throughout the AI Plan. The most prominent, however, is in the context of deregulation of AI itself. Since its first actions on AI in January 2025, the administration has expressed concerns that regulation in the AI space could hinder technological advancement and innovation. The AI Plan recommends various agencies taking a hard look at existing regulation and considering a state’s “AI regulatory climate when making funding decisions”. This latter point is notable, indicating the administration has not given up its efforts to implement a moratorium on state AI regulation. The second most prominent reference to deregulation is in the context of infrastructure development. The AI Plan focuses on the fast-tracking of permitting and additional exclusions under the National Environmental Policy Act to accelerate the build out of new infrastructure, from data centers to the US electrical grid.
The AI Plan does mention a need to increase regulation in certain areas. The policy recommendations suggest additional processes for US AI and semiconductor export controls and enforcement. In addition, with a nod to security, the AI Plan suggests additional security controls over critical infrastructure and certain providers and processes in the realm of biosecurity. Apart from these handful of proposed additional regulations, as for AI itself and the infrastructure needed to support it, “deregulation” is the word of the day.
Investment. Investment recommendations are the most prominent recommendations across all three pillars of the AI Plan. The investment recommendations span a huge swath of investment activities. Additional investment for job training and education in the AI space is a focus of a number of recommendations. This investment includes support for programs and apprenticeships, training opportunities, and the development of a skills framework. Investment in computing power, accessibility to computing power, and the electrical grid supporting that computing power is another key focus of the recommendations. A third significant focus of the recommendations is investment into additional infrastructure to reduce supply chain challenges and support domestic production of components. The policy recommendations also include investment in automated, cloud-enabled labs for research; investment in researchers to build and release datasets; and, of course, investment in the research and standardization efforts discussed below, as well as the talent needed to develop them.
Research and Standardization. Research and standardization are also recurring themes throughout the AI Plan. These recommendations emphasize a need for high-quality data, and recommend policies related to data standards at both domestic and international levels. Proposed standardization measures focus on best practices and standards within regulated industries, such as energy, agriculture, and health care. The administration believes that regulatory uncertainty may be contributing to slower industry adoption of AI. It sees the establishment of best practices and standards as a way to encourage broader industry AI use and engagement.
The AI Plan also proposes evaluation mechanisms and a technology development program to advance AI interpretability, control systems, and adversarial robustness. Additional research recommendations include publishing a new National AI Research and Development (R&D) Strategic Plan, and to evaluate the impact of AI on American workers, including the establishment of an AI Workforce Research Hub.
Cybersecurity. Although the current administration has previously taken measures to reduce cybersecurity obligations related to AI, the AI Plan does acknowledge some of the cybersecurity challenges at play with AI in critical infrastructure sectors as well as the private sector. The AI Plan contains a number of cybersecurity recommendations, such as developer collaboration for cyber-attacks and known vulnerabilities, attention to deepfakes, security of data centers, AI incident response, and the promotion of security-by-design development principles.
Accompanying Executive Orders
The same day the AI Plan was released, the administration issued three separate Executive Orders on AI.
- “Preventing Woke AI in the Federal Government”. This Executive Order directs federal agencies to only procure large language models (LLMs) that adhere to “Unbiased AI Principles” defined in part as “neutral, nonpartisan tools that do not manipulate responses in favor of ideological dogmas such as DEI.” This procurement standard is intended to alleviate the concern that “ideological biases, such as those driven by DEI, can distort AI outputs, undermine historical and scientific accuracy, and erode public trust in AI systems.” It remains to be seen how this will be perceived by the AI industry (which has largely benefited from the administration’s actions on AI thus far) as diversity of data, and the awareness of potential bias within the same, is a foundational component of building trustworthy AI.
- “Accelerating Data Center Infrastructure Development”. This Executive Order takes additional steps towards the administration’s deregulation efforts by streamlining the permitting process for certain AI data centers and infrastructure and financial support for the same. The Executive Order also directs the Environmental Protection Agency (EPA) to expedite the return of Superfund and Brownfield sites for construction of data centers and expressly makes available federal land for data center purposes.
- “Promoting The Export of the American AI Technology Stack”. Looking towards the administration’s goals of promoting US AI and pushing US AI governance standards internationally, this Executive Order directs the Secretary of Commerce to establish the American AI Exports Program. This program is intended to support the development and deployment of full-stack AI packages into the international market by industry-led consortia. Approved participants will be eligible to access support and diplomatic resources, as well as federal funding, to assist in the export and deployment of “full-stack” packages, which must consist of hardware, data systems, AI models, cybersecurity measures, and sector-specific applications.
What to Expect Going Forward
With many policy points in the AI Plan requiring different government agencies to create processes, programs, regulations, and procedures, the landscape for US AI will continue to evolve. Organizations across all sectors and industries should look for additional opportunities for AI development and support from the federal government as we expect additional Executive Orders from the White House in support of the AI Plan.
Special thank you to summer associate, Topher Kittilson, for his assistance with this client alert.
