US President Donald Trump’s sudden shift toward a strict and unpredictable administration of artificial intelligence has some in the industry yearning for some Biden-style regulation.
It’s a change few would have predicted at the start of Trump’s second term, after he swept into office on a wave of donations from Silicon Valley billionaires who warned former President Joe Biden.AI Securitythe policy would destroy American innovation. Trump had signaled a desire to leave AI alone to thrive, and in his first year he was more soaimed at stopping statesdue to controlling technology.
But the arrival of powerful new AI models from Anthropic and OpenAIled to the White Houseto restrict this month the ability of companies to release their most advanced products, for fear that bad actors could use them to launch cyber attacks. The transition to new AI controls has been confusing and imprecise — leaving America’s AI industry in limbo in the meantime.China’s competitors are on the rise.
The unpredictability was therewill be shown again on Fridaywhen the Trump administration partially rescinded its export ban on the most advanced form of Anthropic artificial intelligence – thereby intensifying the confrontation thatcaused chaos in the US AI industry.
But Anthropic’s second high-end model, called the Fable 5, is still under wraps for reasons that remain unclear. And Anthropic’s main competitor, OpenAI,limited the release of its most advanced modelthis week because of similar cyber concerns from the White House.
The moves have hurt an industry that believed it had an ally in the White House, but now faces mounting bipartisan opposition and an uncertain regulatory regime whose scope and scale seem to be changing by the week.
One executive at an AI company, who spoke on condition of anonymity to avoid retaliation, was highly critical of the barriers the administration has placed in the way of new designs.
“This looks like a true European-style licensing regime,” the executive said.
Paul Lekas, head of global public policy and government affairs at the trade group Software & Information Industry Association, which represents some of the leading AI companies, said there is a “real need for a formal process.”
“We want to avoid a situation where the release of any mod or piece of software is dependent on an ad hoc process and a one-time licensing process,” Lekas said.
One OpenAI executive, who spoke on condition of anonymity to describe discussions with the administration, said the industry is seeking “clarity” from the administration — a phrase repeated in interviews with half a dozen other lobbyists and industry representatives.
But industry representatives also said they were wary of pushing the White House for a response, lest they end up being hit with export controls or other blunt regulatory tools.
“It feels like they’re walking on eggshells a little bit,” said one AI policy adviser who works with major frontier labs, speaking on condition of anonymity to clarify the situation.
White House spokeswoman Liz Huston, in a statement, defended the president’s efforts to facilitate innovation in the AI industry, including authorizations to fast-track AI infrastructure and signing an executive order that aimed to end patchwork of different government laws on the technology.
“President Trump has clearly and repeatedly stated his goal: to ensure continued US dominance in AI and other cutting-edge technologies,” Huston said. “President Trump and the entire Administration will continue to strengthen America as the world’s leading innovation power.”
The first sign of the administration’s pivot on AI policy came in early June when, after thatweeks of debateTrumpsign the executive orderwhich set up a voluntary vetting process for companies developing advanced AI designs. But before the order was implemented, the White House went further – in mid-June, it putexport controlon Anthropic’s Mythos 5 and Fable models, preventing the company from releasing its new product due to potential security issues. And this week, ruleit pressured OpenAI to blockrelease of its new GPT-5.6 model to a small group of partners approved by the administration.
Saif Khan, a former emerging technology adviser in the Biden administration, called the Trump administration’s actions an overreaction to predictable security concerns, and suggested it stems from the White House’s previous approach to AI rules.
“Because there has been an aversion to risk, there has been no preparatory work, no hiring of experts that you need to do this work,” Khan said. “And now you have this opaque, vibe-based system for what’s going to be approved and what’s not.”
Khan called the White House’s new approach worse for the AI industry than anything the Biden administration had envisioned, including legislation introduced by the former leftist office that would have imposed export controls to certain countries on semiconductor chips and the weight of advanced AI designs.
“The current administration’s actions have resulted in an almost complete freeze on new releases,” Khan said. “And that will start to really affect the company’s bottom line.”
Dean Ball, former Trump administration official and OpenAI’sthe coming prince of the strategic futureIt said it was inevitable that the White House would need to take a more serious approach, but added that it could find a middle ground.
“It may be true that a holistic view is not appropriate for this technology, and it may also be true that, while the Trump administration’s concerns here are 100 percent correct, there are various ways that I think they are more likely to deal with these legitimate problems,” Ball said.
Despite the growing confusion, some in the AI industry are hopeful that the White House will return to its original hands-off base.
“They are in the middle of the process,” said an OpenAI executive. The person suggested the administration will soon finalize the executive order that Trump signed earlier this month — and that once it does, it could replace the current crackdown with the voluntary vetting regime outlined in the order.
“I think they understand that it’s important to get to the finished system as soon as possible,” the OpenAI executive said. “Because labs continue to produce models, and it’s important for labs to continue to produce models to ensure that America remains a leader in AI.”
Overall, many in the AI industry have agreed that the White House’s new focus on regulation is a necessary change, even if they disagree on the implementation.
“In the big picture, yes, there are things that the administration is doing that I’m not a big fan of, in terms of suddenness and openness and severity, but the bottom line is that I’m glad they’ve come to the conclusion that they have — to take these things seriously,” Ball said.
Lekas also said that the tech industry is creating a “coordinated push for a real system” on advanced AI rules. And he said the tech lobby wants Washington to put it in writing — whether it’s through an executive order or more permanent levers in Congress.
The effort will require technology companies to agree on a standardized approach to AI security. If they cannot row in the same direction, they may continue to receive the same unpredictable treatment.
“It would force the industry to unite in best practice,” Lekas said.
