Dean Ball, the newly appointed head of strategic futures at OpenAI, said he did not like the abruptness, opacity, and strictness of some of the administration’s actions. At the same time, he said he welcomed the government's growing seriousness about AI safety.
Back in 2023, OpenAI CEO Sam Altman warned Congress that the technology could cause serious harm if its development went in the wrong direction. At the time, he said the company wanted to work with the government to prevent such risks.
Anthropic CEO Dario Amodei expressed a similar view. In an essay, he wrote that humanity would find it very difficult to cope with the changes AI will bring, and that success would depend on people’s resolve and responsibility.
The article’s author noted that leaders of major AI companies had warned in advance about potential risks associated with the technology’s development. At the same time, in the author’s view, the American public did not ask those companies themselves to decide the future of AI.
The report also cited polling commissioned by Anthropic. Only 15% of Americans said they trust AI companies to make decisions about how such technologies should be developed and used. Seven in ten respondents opposed building data centers near their homes and said AI development should be slowed. In addition, 87% said foreign countries could use artificial intelligence to attack the United States within the next 20 years.
After taking office, President Donald Trump and Vice President JD Vance signaled reluctance to impose strict AI regulation. Speaking in Paris in February 2025, Vance said excessive rules could slow the development of one of the most promising technologies of the modern era.
Until recently, the administration’s only major regulatory move had been to designate Anthropic as a supply chain risk. The government later effectively paused the release of the company’s new Claude Fable 5 model.
Saif Khan, a former technology adviser in President Joe Biden’s administration, told Politico that the government’s current actions had nearly halted the release of new models and were already beginning to hurt company revenues.
OpenAI is also feeling the effects of the new policy. The company is gradually opening access to GPT 5.6 only to a limited group of users and is in talks with the Trump administration about the procedure for launching it. At the same time, OpenAI stressed that it does not see such a government approval mechanism as a desirable long-term practice.
According to an unnamed adviser on policy for advanced AI models, companies are now acting very cautiously out of fear of new restrictions.
The article also noted that U.S. cybersecurity experts warned after the Fable 5 ban that delays in the United States could give Chinese labs an advantage by allowing them to continue advancing their own AI models more quickly.
Earlier this month, the Trump administration issued an order asking — but not requiring — companies to submit new AI models for federal review. OpenAI said it is working with the government to create a process that would apply to future launches.
The author also noted that the current U.S. regulatory system is not based on separate laws passed by Congress. Instead, decisions about what AI companies are allowed to do depend largely on the position of President Donald Trump.
In conclusion, the author argued that fears voiced by leaders of major AI companies several years ago have already become reality. In the author’s view, the current pause in the development of some models may prove temporary, while the broader regulatory approach may not change significantly.