In recent weeks, a remarkable shift has occurred in the relationship between the artificial intelligence industry and the U.S. government. According to a new report from Politico, executives and policy leaders at major AI companies—often referred to collectively as 'Big AI'—are privately and publicly acknowledging that they needed the Trump administration to set clear rules about what is not acceptable. This admission comes amid a sudden crackdown on Anthropic, a leading AI firm, whose latest model was effectively blocked from release due to national security concerns.
The report paints a picture of an industry that had long operated under the assumption of a deregulatory environment. In early 2025, Vice President JD Vance delivered a speech in Paris declaring that the Trump administration would not regulate AI, arguing that doing so would paralyze a transformative technology and cede advantage to rivals like China. For months, this hands-off stance seemed to give AI companies carte blanche to accelerate development. But that changed abruptly earlier this month when the administration declared Anthropic a supply chain risk and effectively imposed a moratorium on new model releases pending federal review.
The response from Big AI has been mixed. Some executives express frustration at the suddenness and lack of transparency. 'There are things the administration is doing that I'm not so much of a fan of, in terms of the abruptness and the opacity and the strictness, but the more fundamental point is that I'm glad they've arrived to the conclusion that they have — to take this stuff seriously,' Dean Ball, recently hired as OpenAI's Head of Strategic Futures, told Politico. This sentiment echoes a broader realization: the industry's earlier warnings about AI risks may have been prescient, and the lack of guidance left it vulnerable to arbitrary enforcement.
To understand this paradox, it is helpful to recall the rhetorical strategy used by AI leaders in recent years. In 2023, OpenAI CEO Sam Altman testified before Congress, stating, 'I think if this technology goes wrong, it can go quite wrong. And we want to be vocal about that. We want to work with the government to prevent that from happening.' Similarly, Anthropic CEO Dario Amodei wrote in his essay 'The Adolescence of Technology' that humanity's ability to survive the AI-driven turmoil 'will depend on our character and our determination as a species, our spirit and our soul,' adding that 'the years in front of us will be impossibly hard, asking more of us than we think we can give.'
These statements were often dismissed as performative humility or strategic fear-mongering. But the current situation suggests they were also accurate predictions. The companies now find themselves in the painful position they warned about: the government has finally intervened, but not in the predictable, orderly manner they had hoped for. Instead, the Trump administration has used executive orders and national security directives to impose ad hoc restrictions, bypassing the legislative process entirely. This has created an environment where companies are unsure of the rules from one day to the next.
The public, meanwhile, remains deeply skeptical of the entire enterprise. According to a survey conducted by Anthropic itself, only 15% of Americans trust AI companies to make decisions about how AI is developed and used. 7 in 10 oppose data centers in their area, and 87% believe it is likely that foreign governments will use AI to attack the U.S. within the next 20 years. The industry's promise of a technological utopia has not resonated with a populace that sees only risks and few benefits.
The comparison to a dentist is apt. In the original article, the author uses an extended metaphor of a dentist showing his tools to a child before extracting a tooth. The AI companies, like a dentist who arrives unbidden, have displayed their capabilities—the vast language models, the generative agents, the autonomous systems—without a corresponding sense of consent from the public. Now, the administration has finally 'paused the dental procedure,' but the instruments remain on the tray, and the underlying dynamic has not changed.
One of the most striking aspects of the current situation is the geopolitical dimension. The administration's primary concern appears to be preventing China-linked groups from gaining access to frontier models. This was reportedly the trigger for the Anthropic ban: concerns that Fable 5's guardrails could be jailbroken and that Chinese state-linked entities had obtained unauthorized access during a VIP testing phase. Yet, as cybersecurity experts have pointed out, this heavy-handed approach risks stifling U.S. innovation while Chinese labs continue development unhindered. 'Rival labs in China will be able to seize upon the disorder by pushing ahead with their own AI development, while labs in the U.S. get bogged down trying to figure out what is, and what isn't, allowed from them,' noted one observer.
The regulatory landscape is now a patchwork of executive orders, agency directives, and ad hoc governmental pressure. The Trump administration released an Executive Order in early June requesting—but not mandating—that AI companies submit their models for federal vetting. OpenAI, in its blog post about the GPT 5.6 series, claimed to be 'working with the Trump Administration to develop the cyber Executive Order framework and a repeatable process for future model releases.' But the absence of congressional legislation means that the rules are entirely subject to presidential discretion. What is and isn't okay for AI companies to do is currently a matter of whether Donald Trump is pleased with what he is seeing.
This raises profound questions about the future of AI governance. The industry itself has called for clear, stable regulation for years, but it now finds itself in a regulatory vacuum where the only authority is the whims of an executive. The companies are 'walking on eggshells,' as an anonymous policy advisor told Politico. Meanwhile, the public remains largely excluded from the conversation. The only feedback mechanism is the market and occasional protests, but no democratic process has decided how fast or how far AI should develop.
In the background, the AI companies are continuing to release models to select VIP customers, navigating a murky environment where a single security incident could bring a full stop to operations. The tension between the desire for rapid deployment and the need for safety is at an all-time high. The industry's earlier warnings about the pain of this transition now seem less like hyperbole and more like a simple statement of fact.
As the Trump administration decides whether to lift the moratorium and allow Anthropic's Fable 5 or OpenAI's GPT 5.6 to reach the broader market, the fundamental question remains: Who gets to decide what is not okay? The AI companies, the government, or the public? In the absence of a clear answer, the current pause feels less like a resolution and more like a breath held before the next crisis.
Source: Gizmodo News