When President Donald Trump returned to office, he called himself the anti-Biden on AI. First, he halted Biden’s Executive Order 14110, which required “safe, secure, and trustworthy” AI. Then he replaced it with his own directive, “Removing Barriers to American Leadership in AI,” ordering agencies to repeal or weaken rules that were seen as barriers to innovation.
In short, American AI vendors could do whatever they wanted. But that was then. Today is a different story.
While Trump has yet to issue a new AI Executive Order, we do know that his team is preparing an AI task force of tech executives and government officials to oversee AI. Specifically, they are considering requiring all new “high-risk” models to undergo formal government review before they can be used.
And that will be accepted.
What we do know is that National Economic Council Director Kevin Hassett said: “We are potentially looking at an executive order to give a clear roadmap to everyone on how this is going to go and how future AI that potentially creates vulnerabilities would have to go through a process so that they can be released after being proven to be safe – just like an FDA drug.”
Yes, Trump’s “people” will frame this change as a response to escalating cybersecurity and national security risks, not as a broader adoption of EU-style AI regulation. Yes, they are looking at the Anthropic Mythos and its potential use by hackers.
At the same time, they emphasize that they want to avoid “burdensome” controls on everyday AI applications. Cutting-edge models that could enhance cyberwarfare, biological threats or other strategic risks are another matter.
This is a big change from last summer, when Trump Reported: “We need to grow this baby [of Artificial Intelligence] and let it thrive. We can’t stop it. We can’t stop it with politics. We can’t stop it with stupid rules.”
Now he seems to think rules are a good thing. Darrell West, a senior fellow at the Brookings Institution's Center for Technology Innovation, has implied that Trump is returning to Biden's politics. Just don't tell him. He'll have a seizure.
While Trump and his cohorts are still figuring out exactly how they want to govern — sorry, regulate — AI, the Department of Commerce's Center for AI Standards and Innovation (CAISI) has announced new agreements with Google, DeepMind, Microsoft, and xAI. According to these new policy statements, CAISI will conduct assessments pre-development and targeted research to better assess the capabilities of artificial intelligence and advance the state of artificial intelligence security.
CAISI Director Chris Fall said: “Independent, rigorous measurement science is essential to understanding artificial intelligence and its implications for national security.”
You may have noticed that Anthropic wasn't invited to this heated policy meeting. Funny, since most observers believe that the Mythos was the model that broke the “do whatever you want” motto for artificial intelligence in the Trump White House.
This was because the months-long dispute between the government and Anthropic is still ongoing. The Trump team demanded that federal agencies not to use the company's tools and Anthropic is challenging this policy in court..
Recently, however, Trump's tone has softened. Trump told CNBC that Anthropic is “taking shape.” If he can’t make peace with Iran, maybe peace with Anthropic will please him. On the other hand, we also know that Trump’s people are considering banning companies from “interfering” with the government’s use of AI models. Are you hearing that, Anthropic? You’ll follow the boundaries!
Meanwhile, Gregory Falco, assistant professor of mechanical and aerospace engineering at Cornell, pointed out the obvious: “The federal government does not currently have the internal technical expertise, infrastructure, or day-to-day knowledge required to directly evaluate these systems on its own.” Expertise is something that is sorely lacking from the Trump team on all issues.
“At the same time,” Falco says, “a purely voluntary model of self-government is not enough.”
What I think will happen is that the AI companies that play along with Trump will end up “ruling” AI along with some Trump loyalists.
I wouldn't be surprised if one of Trump's goals is not so much to make AI safer as to ensure that the answers it provides are the ones he and his regime want.
Although the press releases will range from very select to rare, I said I'd pass...because sometimes the editors hide.
