Hours after returning to the White House, President Donald Trump made a significant move toward the future of artificial intelligence by revoking former President Joe Biden’s restrictions on the rapidly evolving technology.
However, it is unclear what Trump would do next and how it will differ from his predecessor’s efforts to secure AI technology. The incoming administration did not reply to calls for comment on the overturned Biden policy, and even some of Trump’s most ardent tech sector backers are unsure.
“I think that the previous order had a lot in it,” said Alexandr Wang, CEO of AI startup Scale, describing Biden’s 2023 executive order on AI as unnecessarily long but refusing to specify what was bad. “It’s difficult to remark on any specific aspect of it. There are certain portions of it that we firmly agree with.”
Wang, who flew to Washington to witness Trump’s inauguration, is similarly convinced that brighter days are ahead. He and other Silicon Valley CEOs who previously worked with the Biden administration have welcomed Trump and want to steer his policy toward less constraints.
In its early days, Trump’s team has already “set the tone for a very productive administration with a lot of deep collaboration between industry and government,” Wang stated.
There’s not much left to repeal.
Much of Biden’s directive sparked a rush across federal agencies to investigate AI’s influence on everything from cybersecurity concerns to education, workplaces, and public benefits. That task is finished.
“The reports have been written and the recommendations generated, and they’re available for everyone to build on,” said Alexandra Reeve Givens, CEO of the nonprofit Center for Democracy & Technology. “The executive order’s work is completed, whether or not it’s rescinded.”
According to her, these studies are helping to enlighten the corporate sector, federal authorities, and state governments.
Not only that, but much of the standard-setting provided by Biden’s order followed in the footsteps of earlier AI executive orders issued by Trump during his first term and continued over into the Biden administration.
“If you look past the kind of political positioning on this, the Biden executive order built upon themes that were established in the first Trump administration and have been reiterated by bipartisan voices in Congress,” according to her.
Regulating powerful AI
One crucial part of Biden’s AI directive, which remained in place until Monday, required tech companies developing the most powerful AI models to provide information with the government about how such systems function before they are released to the public.
In many respects, 2023 marked a shift in the AI conversation. ChatGPT was a revolutionary concept, and Elon Musk had previously advocated for a freeze on advanced AI development. Biden’s personal concerns were heightened after viewing Tom Cruise’s film “Mission: Impossible — Dead Reckoning Part One,” in which the planet is endangered by a sentient and rogue computer, according to his then-deputy chief of staff.
The executive order follows public assurances made to the Biden administration by Internet giants such as Amazon, Google, Meta, Microsoft, and OpenAI to support third-party monitoring.
However, the directive went a step further, using the Defense Production Act, which goes back to the Korean War, to require corporations to provide safety test results and other information if their AI systems hit a particular level.
Little is known publicly about how those secret communications really functioned, although several Trump supporters, like venture investor Marc Andreessen, who now serves on the board of Facebook parent Meta Platforms, slammed the government’s investigation last year.
Over the summer, Andreessen expressed concern about “the idea that we’re going to deliberately hamstring ourselves with onerous regulations while the rest of the world and China light up on this.”
Ideological disagreements about AI
Trump is following through on a campaign threat to revoke Biden’s AI directive. His campaign platform defined it as stifling AI progress and putting “Radical Leftwing ideas on the development of this technology,” linking it to Musk and other Trump backers’ larger fears about “woke AI” chatbots with liberal biases.
However, the Biden order itself did not limit free expression. Some rules aimed to establish standards for watermarking AI-generated content as part of an effort to decrease the risks of impersonation and abusive sexual deepfakes. The decision also instructed numerous federal agencies to protect against possible AI-related abuses, warning against reckless usage that “reproduced and intensified existing inequities, caused new types of harmful discrimination, and exacerbated online and physical harms.”
One former White House science advisor who helped create Biden’s rights-based AI strategy called Trump’s decision a “politically motivated repeal with no thoughtful replacement.”
Trump’s move indicates that he is “less supportive than the Biden administration of issues around privacy, people’s civil liberties and civil rights, and just concerns around safety more broadly with regards to advanced systems,” according to Alondra Nelson, the former acting director of the White House Office of Science and Technology Policy.
Addressing these issues is critical for consumers to use the AI tools that corporations are developing, said Nelson, who is now a fellow at the Center for American Progress.
“Americans have some of the highest rates of mistrust of AI in the developed world,” she explained, citing polls.
Pivot to AI common ground.
Some of Biden’s AI initiatives, such as the year-old AI Safety Institute focused on national security, remain in place, at least for the time being. Trump has also yet to comment on Biden’s larger fight with the tech industry, which involves pending rules barring AI chip shipments to more than 100 nations in an effort to offset China’s backdoor access to them in regions such as the Persian Gulf and Southeast Asia.
Nor has Trump reversed Biden’s most recent AI executive order, which was issued a week ago and aims to eliminate barriers to the construction of AI data centers in the United States while simultaneously encouraging the data centers to use renewable energy.
On Tuesday, Trump touted a joint venture called Stargate, created by ChatGPT manufacturer OpenAI, Oracle, and SoftBank, that will spend up to $500 billion in AI data centers and the infrastructure to support them. At a news conference, he appeared unfamiliar with Biden’s recent AI directive but stated that he would not withdraw it.
“That sounds to me like it’s something that I would like,” Mr. Trump remarked. “I would want to see federal lands open up for data centers. I believe they’ll be really important.”