“Trump’s executive action on AI rolls back Biden-era attempts to tightly monitor and control the design and governance of new models,” said Daniel Cochrane, a tech policy associate at the Heritage Foundation. “President Trump’s executive actions are a good start to rolling back that dangerous framework. They also clear a path for unleashing American AI innovation.”

Former President Joe Biden tried to singlehandedly regulate AI by issuing Executive Order 14110 on Oct. 30, 2023. Titled “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence,” it required AI companies to share key safety information and test results with the federal government before new models were released to the public.

All of that is now off the table.

“To commence the policies that will make our nation united, fair, safe, and prosperous again, it is the policy of the United States to restore common sense to the federal government and unleash the potential of the American citizen,” reads Trump’s Day One executive order rescinding more than 60 of Biden’s decrees, including those relating to artificial intelligence.

The tech industry always opposed Biden’s attempts to control it, and it now enjoys freedom from his restrictions. Trump issued another executive order on Jan. 23, titled “Removing Barriers to American Leadership in Artificial Intelligence,” that further rolled back regulations.

“The United States has long been at the forefront of AI innovation, driven by the strength of our free markets, world-class research institutions, and entrepreneurial spirit,” the order reads. “To maintain this leadership, we must develop AI systems that are free from ideological bias or engineered social agendas.”

The order revokes Biden’s policies that Trump says “act as barriers to American AI innovation, clearing a path for the United States to act decisively to retain global leadership in artificial intelligence.”

The next week, the urgency of that message was underscored by the emergence of DeepSeek, a high-performing AI model developed in China that sent U.S. tech stocks into a broad selloff. Trump has already announced his own AI initiative, Stargate, touting $500 billion in investment that the president says will be used to build artificial intelligence infrastructure and create 100,000 domestic jobs in the process.

However, despite AI’s promise, some advocates, including Biden, have argued that regulation is necessary to protect against the technology’s dangers, including from discrimination on the part of AI models.

“To realize the promise of AI and avoid the risk, we need to govern this technology,” Biden said when he announced his regulatory executive order. Along with the reporting requirements, Biden’s order created the U.S. AI Safety Institute within the Commerce Department.

Howard Lutnick, Trump’s nominee for commerce secretary, was questioned about the AI Safety Institute during his confirmation hearings. He said the department has the gold standard for cybersecurity and that “AI standards along the lines of that gold standard, that same model, I think, will be very effective.”

“If you think of it as standards, I think we can get bipartisan agreement,” Sen. Brian Schatz (D-HI) said.

“We need to protect America,” Lutnick agreed. “But we also need to make sure it is an American-driven AI model in the world that’s important to us as Americans.”

Cochrane argues that Biden’s regulatory model was deeply ideological and meant, in practice, that government agencies had closed-door access to the most powerful AI models, allowing them to pressure developers into implementing “fixes” ahead of their release.

“It was a blatant attempt to encode censorship into AI from the start,” he said. Cochrane holds that overly aggressive efforts to combat bias can hurt the accuracy of AI, citing how Google’s Gemini created inaccurately racially and ethnically diverse historical figures and images of black Nazis.

“Big Tech effectively sacrificed substantive reliability and quality in pursuit of other political goals, like appeasing the Biden administration and their powerful friends,” Cochrane said.

With Trump’s rescission of the Biden order, there is no longer a regulatory framework for AI.

Proponents of regulation say that some AI tools have been found to discriminate based on race, sex, or disability, with the Associated Press reporting that some medical diagnosis chatbots have been caught “spouting false information” and facial recognition technology has been tied to the wrongful arrests of black men.

Doug Calidas, vice president of government affairs at Americans for Responsible Innovation, acknowledged that such problems exist, though he said companies are strongly incentivized to correct problems on their own irrespective of government decrees.

AI discrimination “is rarely due to any bad intent on the part of the developer, and stems mostly from the fact that capable AI systems must use massive amounts of historical data,” Calidas said. “It’s not a good look for a serious company to release or deploy an AI tool that discriminates in this way, so developers and deployers work to try and remove these biases from their tools.”

Calidas noted that standardized federal standards can help companies remove biases “more effectively for less money.”

Trump’s order does not mean that AI will be completely unregulated forever or even for the duration of his second term in the White House. The order calls for a new AI framework to be developed over the next six months.

Other parts of the federal government are looking into regulation as well. The Senate discussed AI at length in 2023 and 2024, eventually producing a road map for policy priorities. No major legislation has worked its way through Congress to this point, although Senate Minority Leader Chuck Schumer (D-NY) has sought to revive the bipartisan effort this year.

Europe passed a comprehensive regulatory framework for AI last year, though Trump is unlikely to follow the lead of continental legislators on this or other issues.

Samuel Hammond, the chief economist at the Foundation for American Innovation, said Biden’s AI executive order inhibited experimentation while creating a “litany of vague and burdensome” rules. He expects the new framework the Trump administration eventually creates to be better.

Hammond also stressed that some government oversight remains in place, including the AI Safety Institute, which is part of the Commerce Department’s National Institute of Standards and Technology.

“My view is that regulating AI as a category is premature but having some basic monitoring and oversight capacity for the top AI labs is essential for national security,” Hammond said. “Trump’s EO calls for an AI action plan to be developed within 180 days. We shall see what approach they end up taking.”