The Trump administration published its much-anticipated AI Action Plan on Wednesday, a document that takes a sharp shift away from former President Biden’s cautious approach to addressing the risks of AI, and instead barrels ahead with plans to build out AI infrastructure, cut red tape for tech companies, shore up national security, and compete with China.
The downstream effects of this shift will likely ripple throughout various industries and may even be felt by the average American consumer. For instance, the AI Action Plan downplays efforts to mitigate possible harms of AI and instead prioritizes building out data centers to power the AI industry, even if it means using federal lands or keeping them powered during critical energy grid periods.
Much of its effects, however, will depend on how the AI Action Plan is executed, and many of those details have yet to be sorted. The AI Action Plan is more blueprint for action than a step-by-step instruction book. But the direction is clear: Progress is king.
The Trump administration positions this as the only way to “usher in a new golden age of human flourishing.” Its goal is to convince the American public that spending billions of taxpayer dollars on building data centers is in their best interest. Parts of the plan also include policy suggestions around upskilling workers and partnering with local governments to create jobs related to working in data centers.
“To secure our future, we must harness the full power of American innovation,” Trump said in a statement. “To do that, we will continue to reject radical climate dogma and bureaucratic red tape, as the Administration has done since Inauguration Day. Simply put, we need to ‘Build, Baby, Build!’”
The AI Action Plan is authored by the Trump administration’s team of technology and AI specialists, many of whom come from Silicon Valley companies. This includes Office of Science and Technology Policy director Michael Kratsios; AI and crypto czar David Sacks; and assistant to the president for national security affairs Marco Rubio. More than 10,000 interest groups submitted public comments that were considered for the plan.
Deregulation and bringing back the AI moratorium
At the start of this month, the Senate removed a controversial provision in the budget bill that would bar states from regulating AI for 10 years. That provision, if it had been included in the bill, would tie states’ federal broadband funding to compliance with the moratorium.
Techcrunch event
San Francisco
|
October 27-29, 2025
It appears that matter hasn’t been put to rest yet, as the AI Action Plan explores a new way to hinder states from regulating AI. As part of a broader mission to “unleash prosperity through deregulation,” the administration threatens to limit states’ federal funding based on their AI regulations.
The plan also directs the Federal Communications Commission to “evaluate whether state AI regulations interfere with the agency’s ability to carry out its obligations and authorities.” In other words, if state AI regulations touch on radio, TV, and internet — which many do — then the FCC can get involved.
On the federal level, the action plan directs the Office of Science and Technology Policy to ask businesses and the public about any current federal regulations that hinder AI innovation and adoption so that federal agencies can take appropriate action.
Cutting red tape around data centers
Trump’s call for deregulation extends to how the administration hopes to accelerate the buildout of AI-related infrastructure, like data centers, semiconductor fabs, and power sources. The administration argues that existing environmental regulations — like NEPA, the Clean Air Act, and the Clean Water Act — are hindering America’s need to meet the rapid demands of the AI arms race.
That’s why Trump’s AI Action Plan places an emphasis on stabilizing America’s energy grid. At the same time, the plan asks the federal government to find new ways to ensure large power consumers — such as AI companies — can manage their power consumption during critical grid periods.
Certain companies, like xAI and Meta, have been criticized for concentrating pollution in vulnerable communities. Critics have accused xAI of bypassing environmental safeguards and exposing residents to harmful emissions from gas-powered turbines with its Memphis data center.
The action plan calls for creating categorical exclusions, streamlining permitting processes, and expanding the use of fast-track programs like FAST-41 to make it easier for companies to build critical AI infrastructure, especially on federal lands, which includes national parks, federally protected wilderness areas, and military bases.
Tying back to other Trump themes of beating China, the strategy focuses on locking out foreign tech and emphasizing security protections to keep “adversarial technology” — like Chinese-made chips and hardware — out of the U.S. supply chain.
Trump’s war on “biased AI”
One of the main standouts in Trump’s AI Action plan is a focus on protecting free speech and “American values,” in part by eliminating references to misinformation, DEI, and climate change from federal risk-assessment frameworks.
“It is essential that these systems be built from the ground up with freedom of speech and expression in mind, and that U.S. government policy does not interfere with that objective,” the plan reads. “We must ensure that free speech flourishes in the era of AI and that AI procured by the Federal government objectively reflects truth rather than social engineering agendas.”
Despite the intention to ensure that government policy doesn’t interfere with freedom of speech, the AI Action Plan has the potential to do just that.
One of the recommended policy actions is to update federal procurement guidelines to ensure the government only contracts with frontier large language model developers who “ensure their systems are objective and free from top-down ideological bias.”
That language is similar to what The Wall Street Journal reported would be in Trump’s executive order, which is expected to be released later today.
The problem is that achieving objectivity is hard, and the government has not yet defined how it plans to evaluate models on the basis of neutrality.
“The only way to be neutral would be literal non-engagement,” Rumman Chowdhury, a data scientist, CEO of the tech nonprofit Humane Intelligence, and former U.S. science envoy for AI, told TechCrunch.
Anthropic, xAI, Google, and OpenAI have all secured government contracts worth up to $200 million each to help integrate AI applications into the Department of Defense. The implications of Trump’s policy suggestion and his impending executive order, could be far-reaching.
“For instance, an order that says, ‘We won’t do any business, as to AI models or otherwise, with any company that produces a non-neutral AI model’ would likely violate the First Amendment,” Eugene Volokh, an American legal scholar who specializes in First Amendment and Second Amendment issues, said in an email. “An order that says, ‘We will only enter into contracts to buy models that are sufficiently neutral’ would be more constitutionally defensible, though of course implementing it effectively may be very difficult (partly because it’s so hard to know what’s ‘neutral’ in these situations).”
He added: “If the order instructs agencies to select AIs based on a combination of accuracy and neutrality, leaving each agency with some latitude to decide what that means, that might be more viable.”
Encouraging an open approach to AI
Trump’s AI Action Plan aims to encourage the development and adoption of open AI models, which are free to download online and are created with American values in mind. This largely seems to be a reaction to the rise of open AI models from Chinese AI labs, including DeepSeek and Alibaba’s Qwen.
As part of his plan, Trump wants to ensure that startups and researchers working on open models have access to large computing clusters. These resources are expensive and typically were only possible for tech companies that could strike million- or billion-dollar contracts with cloud providers.
Trump also says he wants to partner with leading AI model developers to increase the research community’s access to private AI models and data.
American AI companies and organizations that have taken an open approach — including Meta, AI2, and Hugging Face — could benefit from Trump’s embrace of open AI.
AI safety and security
Trump’s AI Action Plan includes some provisions to satisfy the AI safety community. One of those efforts includes launching a federal technological development program to research AI interpretability, AI control systems, and adversarial robustness.
Trump’s plan also instructs federal agencies, including the Department of Defense and Department of Energy, to host hackathons to test its AI systems for security vulnerabilities.
Trump’s plan also acknowledges the risks of AI systems to contribute to cyberattacks, as well as the development of chemical and biological weapons. The plan asks frontier AI model developers to work with federal agencies to evaluate these risks and asks how they could jeopardize America’s national security.
Compared to Biden’s AI executive order, Trump’s plan puts less of a focus on requiring leading AI model developers to report safety and security standards. Many tech companies claim safety and security reporting is an “onerous” task, which Trump seems to want to limit.
Limiting China
Perhaps unsurprisingly, Trump is bringing his war on China into the AI race with his action plan. A large part of Trump’s AI Action Plan focuses on preventing “national security” threats from accessing advanced AI technology.
Under Trump’s plan, federal agencies will work together to collect intelligence on foreign frontier AI projects that could threaten American national security. In one of those efforts, the Department of Commerce is tasked with evaluating Chinese AI models for alignment with Chinese Communist Party talking points and censorship.
These groups will also conduct assessments on the level of AI adoption among America’s adversaries.
National security
“National security” was mentioned 23 times in the AI Action Plan — more than “data centers,” “jobs,” “science,” and other key terms. The plan’s national security strategy is centered on integrating AI into the U.S. defense and intelligence apparatus, and even building out AI data centers for the DoD, while guarding against foreign threats.
Among other things, the plan calls for the DoD and intelligence community to regularly assess how AI adoption in the U.S. compares to rivals like China and adapt accordingly, and to evaluate risks posed by both domestic and adversary AI systems.
Within the DoD itself, the strategy emphasizes upskilling the military workforce, automating workflows, and securing preferential access to compute resources during national emergencies.
Got a sensitive tip or confidential documents? We’re reporting on the inner workings of the AI industry — from the companies shaping its future to the people impacted by their decisions. Reach out to Rebecca Bellan at rebecca.bellan@techcrunch.com and Maxwell Zeff at maxwell.zeff@techcrunch.com. For secure communication, you can contact us via Signal at @rebeccabellan.491 and @mzeff.88.
