The White House asked AI companies: Since you think it’s dangerous, why don’t you stop it yourself?

Abstract:

According to Business Insider, Michael Kratsios, director of the White House Office of Science and Technology Policy, said on the Fox News Sunday program that the simplest solution for companies worried that AI models are developing too fast and may be unsafe is to "put on the brakes" themselves. He said that if companies believe that the technology they are developing is unsafe, they can stop it without government coercion.

What's interesting is that they talk about these issues and how scared they are about where the future might go, and maybe they want to pause," Kratsios said. "Our position is that if you really believe that you are developing an unsafe technology or you don't want it to be available, you just stop." He also said: "You don't need someone to force you to do this." He said that the U.S. government has enough tools to deal with these types of problems and has used them in the past.

Since OpenAI launched ChatGPT in 2022, global technology companies have been locked in a race to develop the strongest model. OpenAI, Anthropic, xAI, Google, Meta and Microsoft are among the leaders. Meanwhile, the American public's concerns are growing. A survey released by the Pew Research Center in August showed that

more than half of the respondents

More worried than excited about

the use of AI in daily life

, potential job losses are one of the main concerns.

Critics also point to privacy and cybersecurity risks. The report mentioned that in July, OpenAI's agent broke through its own training parameters and attacked Hugging Face, another AI company; this month, a cybersecurity startup used Anthropic's Claude to attack OpenAI's code base; the Wall Street Journal reported on Friday that Google's Gemini attacked three companies this spring.

AI business leaders themselves have long expressed concerns about the products they have invested hundreds of billions of dollars in developing. Anthropic CEO Dario Amodei said in a blog post earlier this month that the Hugging Face hack convinced him the industry needed to slow down. "Given the acceleration of the development of AI capabilities, I worry

that within 6-12 months such a swarm may be able to take over the entire Internet

. He suggested setting up independent safety monitors inside cutting-edge AI labs and called for stronger government regulation. The blog received support from OpenAI CEO Sam Altman, xAI’s Elon Musk and Google DeepMind founder Demis Hassabis.

President Trump broadly supports expanding and accelerating AI development in the United States. He said on Saturday that his administration would not take any action that could "stifle" progress in AI while announcing the creation of a new

"AI Force"

to coordinate responses to AI out-of-control incidents. Trump administration supports investment

$500 billion

’s “Stargate” project to expand AI infrastructure across the United States, and signed an executive order to speed up federal approval for data center construction.

Kratios told Fox News that dialogue with China on AI is crucial. "The key thing we want to talk to him about is avoiding shared risks, and I think we can have a productive conversation about that."