For months now, the White House has hinted that it may try to regulate the AI industry. Just two weeks ago, the nation’s top tech executives—including Sam Altman and Dario Amodei—were invited to attend a signing ceremony for a long-awaited executive order on AI. But a few hours before the ceremony, Donald Trump canceled it. America is leading the world in the AI race, the president told reporters at the time, “and I don’t want to do anything to stop that lead.”
Apparently, Trump has changed his mind again. Earlier today, President signature executive order that would create a process for major AI companies to voluntarily share certain upcoming designs with the government for security testing up to a month before a wider release. OpenAI, Anthropic, and others will also be asked to work with governments to improve federal, state, and local cyber defenses. White House spokeswoman Liz Huston told us that the policy reflects “a common way of working with industry to balance innovation and security.”
The order itself is toothless: Even before today, major AI companies already had agreements that allowed governments to test their models for security risks. The new law “effectively formalizes what has already been happening between the US government and AI companies,” Daniel Remler, an AI expert at the Center for a New American Security, told us.
But the executive order makes sense since the president is doing something—anything—about AI. At the start of his second term, Trump signaled to tech companies that he would stay out of the way. Last January, he rescinded a set of Joe Biden-era policies, calling the rules “dangerous” and “an obstacle” to America’s AI leadership. Even the preamble to today’s executive order celebrates that Trump “triggered massive technological growth” by “reducing the bureaucratic restrictions that the previous administration placed on America’s AI developers.” Yet key parts of those allegedly dangerous AI principles of the Biden era—voluntary contracts sharing information about advanced AI designs with federal agencies, for example, as well as federal programs to augment AI for cyber defense—are very similar to today’s AI executive order. Dean Ball, a former AI adviser to the Trump administration, he wrote that the policy is “more intrusive” than Biden’s executive order.
Today’s order could still be stronger. When the White House began reviewing possible regulatory measures in May, one administration official he suggested that AI models would be reviewed “just like FDA medicine.” Even the leaked text of the deal Trump had planned to sign last month would have been more burdensome for tech companies. After David Sacks, White House the former king of AIreportedly called the president to complain, Trump canceled the signing ceremony. Today, after the new order was announced, Sacks declared the provisions a “game changer” at X—despite the fact that the new government review process is not much different from what he had previously opposed. This means that the two former White House AI advisers—Ball and Sacks—disagree on whether this order is a good one.
Meanwhile, joining Sacks in praising the law is Steve Bannon, Trump’s former chief strategist and a leading critic of AI on the right. “It’s not perfect,” he told us. “But in terms of direction, it’s very good.” As Bannon sees it, despite the fact that the order is weaker than previous versions, the marking rules are a step in the right direction.
The whole, chaotic saga—the White House, the confused statements from Trump’s top spokespeople and elites—is just the latest in a long line of strange, often contradictory AI-policy positions. Trump’s approach to AI has been inconsistent, if not consistent, almost since the day he took office. Consider that, for all the talk of cyber security, this administration also has intestines Cybersecurity and Infrastructure Agency, a government agency that aims to protect the nation from hackers. CISA also happens to be one of the main federal agencies tasked with implementing today’s executive order.
Or take the White House’s relationship with Anthropic. On the one hand, Anthropic likely initiated the executive order in the first place. In April, the company announced the Claude Mythos Preview, a new model with high hacking capabilities that it has anxiety flared up on the growing power of AI companies. Since then, the president has been seen as favorable to Anthropic. Dario Amodei, the company’s CEO, visited the White House that same month for talks on the future of the government’s relationship with the company. “I like people with a high IQ, and of course they have a high IQ,” Trump later on he told reporters about Anthropic leadership.
On the other hand, the Trump administration appears to be fighting in court to prevent Anthropic from doing most of the national security work. In February, the Pentagon designated Anthropic a “supply chain risk” after a high-profile contract dispute over the use of AI in warfare, essentially declaring it a national security risk for the military to even touch Anthropic’s products. At the end of April, when Anthropic tried to give access to Mythos to more cyber security companies – according to today’s executive order – the White House. it seems have, inexplicably, prevented action. (An Anthropic spokesperson pointed us to a post on X that the company to be called today’s executive order “an important step in strengthening America’s leadership in AI.”)
Then there are the administration’s attitudes toward China. Trump has repeatedly emphasized the need to deregulate the AI industry to stay ahead of China. At the same time, he has also allowed Nvidia to sell some of its most advanced AI chips to Chinese companies, removing export controls imposed by the Biden administration on Chinese AI development. (Anthropic, by the way, rejected a Chinese professional institute to reach Mythos.) Trump, in the name of beating China, has pushed to remove regulatory restrictions on data center construction: “Build, baby, build,” he said last July. But when the uproar arose about data centers raising electricity bills, the White House announced a voluntary commitment for AI companies to take several steps that would prevent everyday people from paying for data center electricity. Build, baby, but wisely.
Indeed, at least some of the changes seem to be driven by public opinion. In the past several months, as AI models have done improvedattitudes towards technology have get upset. Today’s order allows the administration to appear as if it is taking stronger control of AI—but it doesn’t require industry to do much, if anything. Trump is trying to score points with the public and Silicon Valley. But in doing so, he doesn’t say or do anything fundamental at all.
The use of AI is burning the US economy, people fear losing their jobs to AI, and communities across the country are rallying against data centers. Different political figures like Bannon and Bernie Sanders express their concern about AI and the concentration of power among industry executives. This may seem like an open call to the president of the United States, and the most popular person at the time. Instead, the White House spent weeks drafting an executive order that relies on the voluntary cooperation of the AI industry. With Anthropic, OpenAI, and their competitors becoming major economic and geopolitical powers, the window for one government to seriously regulate AI is closing fast. I hope, it’s not over yet.
