White House Releases Trump’s ‘AI Action Plan’—Here’s What’s in It

The White House on Wednesday unveiled a national strategy to cement U.S. dominance in artificial intelligence—drawing criticism from civil rights and policy groups who say it sacrifices safety for power.

The “AI Action Plan” outlines the Trump administration’s push to fast-track AI infrastructure, export American-made systems to allies, and roll back Biden-era regulations seen as obstacles to growth.

The plan follows Trump’s earlier launch of the $500 billion Stargate Project—a private-sector initiative led by OpenAI, Oracle, and Japanese conglomerate SoftBank to boost U.S. AI infrastructure and growth.

“This plan galvanizes federal efforts to turbocharge our innovation capacity, build cutting-edge infrastructure, and lead globally, ensuring that American workers and families thrive in the AI era. We are moving with urgency to make this vision a reality,” White House Office of Science and Technology Policy Director Michael Kratsios said in a statement.

Key policies in the AI Action Plan include:

• Exporting U.S. AI: The Commerce and State Departments will coordinate with industry to deliver secure, full-stack AI systems—including hardware, software, and standards—to allied nations.
• Accelerating Buildout: The plan fast-tracks permits for data centers and chip fabricators, and launches programs to train high-demand workers like electricians and HVAC techs.
• Cutting Regulations: It removes federal rules that slow AI development and seeks industry input on additional rollbacks.
• Free Speech in AI: Federal contracts will require that AI models remain free from “government-imposed ideological bias.”

Matthew Mittelsteadt, a Technology Policy Research Fellow at the Cato Institute, a libertarian think tank, called the plan a “mixed bag,” praising its emphasis on rapid development compared to Biden-era orders, but warned of political overreach.

“One theme that I think is a positive is the renewed emphasis on innovation,” Mittelsteadt told Decrypt. “The point of this technology is to help people—and that really should be the point of these orders.”

However, Mittelsteadt warned that tying federal contracts to government-aligned models risks politicizing AI development.

“That’s how you twist this space,” he said. “Developers will start building to appease the administration.”

If American AI models begin reflecting political agendas, then foreign users may see them as tools of U.S. influence rather than a neutral technology. That could be devastating to U.S. credibility abroad, he said.

“Anyone who wants to buy American models abroad will view our models as influenced by the US government, just as we view China’s models as influenced by the Chinese government,” Mittelsteadt said. “No one’s going to want to buy these things if they’re seen as tools of Donald Trump, and that is such a mistake.”

AI Action Plan also threatens to withhold federal AI funding from states with “burdensome” rules—a move critics say could chill local innovation and create new constitutional fights.

“I have no clue how they’re going to define ‘burdensome,’” Mittelsteadt said. “States aren’t going to just sit there guessing what qualifies. They’ll keep passing laws, and then we’ll have messy legal battles.”

Mittelsteadt also cast doubt on the administration’s ability to follow through on such threats, especially against tech-heavy states like California.

“The administration talks about cutting off funding, but much of that funding goes to federal entities—like national labs—that are physically located in states like California,” he said. “Are they going to defund a major lab over a state policy? There are a lot of question marks here. A lot of implementation would be inherently messy and, I think, inherently politicized.”

Eric Null, the co-director of the Privacy & Data Project at Center for Democracy and Technology, echoed that concern, adding that the AI Action Plan could potentially undermine effective AI regulation.

“A lot of it is going to be harmful and counterproductive in ensuring AI is developed and used in trustworthy ways,” Null told Decrypt. “For the FCC to carry out this directive, it would have to claim authority over AI in some form, which would be beyond its scope.”

Null also pointed to the policy’s call for AI platforms to be “free from top-down ideological bias,” arguing that it leaves room for partisan interpretation.

“Ideological bias is hard to eliminate from AI, making enforcement tricky. While the rule is meant to apply across political lines, it’s up to the government to decide what counts as bias,” he said. “In the hands of a partisan administration, that could mean uneven enforcement—labeling some views biased while excusing others. I’m not confident it’ll be applied fairly.”