On July 23, the Trump administration released a new artificial intelligence policy blueprint called “Winning the Race: America’s AI Action Plan.” This plan follows a public process that began in February and received more than 10,000 responses. According to an analysis from the R Street Institute, the Action Plan aims to encourage competition and investment in AI while strengthening national interests.
The plan continues a market-focused approach that has characterized Trump administration policies since returning to office. Additional steps are expected as part of this agenda. However, one of these steps has raised concerns about possible government overreach.
President Donald J. Trump recently signed an executive order titled “Preventing Woke AI in the Federal Government.” The order recommends removing terms such as “misinformation,” “Diversity, Equity, and Inclusion [DEI],” and “climate change” from the National Institute of Standards and Technology (NIST) AI Risk Management Framework (AI RMF). It also calls for companies contracting with the federal government on large language models (LLMs) to be free from what it describes as “top-down ideological bias” and to follow principles described as “truth-seeking” and “ideological neutrality.”
The goals of neutrality and fairness in technology are seen as positive but difficult to translate into clear policy. Some experts warn that efforts by government agencies to regulate speech within AI systems—even through procurement—could raise First Amendment issues if taken too far. There is also concern that changing definitions between administrations could create confusion or lead different parties to target various types of speech.
Additionally, some observers argue these efforts might distract from other significant developments at the state level. Several states have considered or passed laws imposing DEI-inspired regulations on algorithmic systems used by private companies.
For example, Colorado enacted SB24-205 (“Consumer Protections for Artificial Intelligence”) in May 2024. The law will take effect in early 2026 and is designed to prevent algorithmic discrimination through preemptive rules governing certain AI systems (https://leg.colorado.gov/bills/sb24-205). A coalition known as the Multistate AI Policymaker Working Group (MAP-WG), which includes lawmakers from more than 45 states, has pushed for similar legislation.
These state initiatives often reference concepts found in President Biden’s earlier Blueprint for an AI Bill of Rights, which called for proactive measures against algorithmic discrimination based on equity assessments (https://www.whitehouse.gov/ostp/ai-bill-of-rights/).
According to Greg Lukianoff, President and CEO of the Foundation for Individual Rights and Expression: “[H]aving the government decide what those principles mean in practice is where things always get hairy.”
The new executive order does not directly regulate how LLMs are developed or used outside federal contracts. However, it gives broad discretion to future administrations regarding how terms like “ideological bias” are defined when awarding contracts.
There are practical challenges associated with implementing these requirements. The executive order notes that guidance from the Office of Management and Budget should consider technical limitations faced by vendors complying with its standards while allowing flexibility so companies can innovate without undue restrictions.
Some analysts worry this approach could impose extra costs on developers who must navigate new compliance requirements related to speech content within their products.
A further concern is whether federal actions might overshadow efforts needed at other levels of government. Unlike federal contracting rules, recent state-level bills would apply directly to private sector development and use of AI technologies across entire markets rather than just within public procurement.
The Trump administration’s Action Plan instructs agencies distributing discretionary funding for AI projects to consider each state’s regulatory climate before awarding funds—a move intended both as leverage against adopting stricter local rules and as a way to protect national priorities related to technological innovation.
Congress has debated proposals such as an AI regulatory moratorium—which would have paused new state-level regulations for ten years—but so far only limited progress has been made toward enacting such measures nationwide (https://www.congress.gov/bill/118th-congress/house-bill/7521/text).
During remarks announcing his administration’s plans, President Trump said: “We need one commonsense federal standard that supersedes all states.” Supporters argue a unified national approach could prevent conflicting mandates that complicate industry compliance while maintaining U.S. leadership in artificial intelligence research.
