White House AI plan places scrutiny on state AI laws

AI Action Plan targets regulations

The plan zeroes in on regulation -- from federal agency policies and rules to U.S. state laws.

The administration asks that the Office of Science and Technology Policy send out a request for information to businesses and the public about current federal regulations hindering AI innovation and adoption. The Office of Management and Budget (OMB) has been designated to work with federal agencies to revise and repeal regulations, policy statements and other measures that hinder AI development and deployment.

The plan also targets U.S. states that have adopted AI laws. Under the plan, OMB is responsible for working with federal agencies to assess AI-related discretionary funding programs. When making funding decisions, OMB and federal agencies will be required to limit funding if states' AI regulatory regimes may "hinder the effectiveness of that funding or award."

Nonprofit consumer organization Consumer Reports issued a statement raising concerns with the White House plan. Grace Gedye, Consumer Reports' policy analyst for AI issues, said in a statement that the AI plan creates confusion for states, including about which federal funds are on the line if their AI law is considered overly burdensome.

Indeed, Valente said the AI Action Plan lacks definitions, making it unclear what state AI laws will be considered unduly restrictive to innovation.

The Federal Trade Commission will also be under review as part of the plan, to ensure investigations commenced under former President Joe Biden "do not advance theories of liability that unduly burden AI innovation."

Jay Timmons, president and CEO of the National Association of Manufacturers, said in a statement that the organization has been advocating for a "pro-AI policy environment," which he said the AI Action Plan accomplishes.

The plan reflects many of the ideas the manufacturers' group has proposed, including permitting reforms to maintain U.S. energy dominance, a more targeted regulatory approach, supporting workforce development and enabling small and medium-sized manufacturers to access AI technology, according to the statement.

Protecting 'free speech'

Ensuring AI models protect freedom of speech and expression is a pivotal concern for the Trump administration.

The AI plan recommends that the commerce department, through the National Institute of Standards and Technology, revise the NIST AI Risk Management Framework. The administration wants to remove references to misinformation, diversity, equity and inclusion, and climate change. The commerce department and NIST will also be researching and evaluating frontier AI models from China.

The industry consensus is that AI is only as good as the way it's trained, which often includes large swaths of varied information and data, said Bonnie Montano, a professor at Georgetown University's McDonough School of Business.

"If people building these models start picking and choosing what to include or exclude, that will certainly cause bias that they say they're trying to avoid," she said.

The White House plan mentions updating federal procurement guidelines to ensure the federal government contracts with frontier large language model developers who keep their models free from "top-down ideological bias."

Valente said the recommendation on ideological biases raises questions for developers on what to do if they incorporate such data and terms into their AI models.

"Does that mean they're being restricted from selling to the federal government?" Valente said.

Building out AI infrastructure and exporting AI products

The AI Action Plan wants to streamline permitting for AI data centers, energy infrastructure and semiconductor manufacturing facilities.

To accomplish this, the plan recommends establishing new exclusions under the National Environmental Policy Act to cover data center-related actions that don't typically have a significant environmental effect. It also recommends expediting environmental permitting by reducing regulations within the Clean Air Act, the Clean Water Act and other related laws.

The plan also recommends making federal lands available to build AI data centers and energy infrastructure.

Protecting U.S. AI infrastructure will mean ensuring the AI computing stack is building on U.S. products, according to the plan. The plan recommends building infrastructure free from foreign adversary information and communications technology and services. It also mentions it will be crucial for the federal government to prevent bad actors from misusing or stealing advanced technologies, as well as monitoring for unforeseen AI risks.

Keeping the U.S. in the AI race also means being able to export AI products. The AI plan says the U.S. must meet global demand for AI by exporting AI hardware, models, software, applications and standards to countries willing to join the U.S. AI alliance.

The plan also recommends that the commerce department establish a program to gather proposals from the AI industry for full-stack AI export packages. Multiple U.S. departments and agencies will then coordinate with the department to expedite deals meeting U.S. security requirements and standards.

While the AI Action Plan includes recommendations to enable AI exports, it also includes a recommendation for U.S. officials to consider new export controls on semiconductor subsystems. The commerce department currently imposes export controls on major systems needed for chip manufacturing, but not the component subsystems, according to the plan.  

Trump was expected to speak about federal policy for AI later Wednesday.

Makenzie Holland is a senior news writer covering big tech and federal regulation. Prior to joining Informa TechTarget, she was a general assignment reporter for the Wilmington StarNews and a crime and education reporter at the Wabash Plain Dealer.