Trump Calls AI Safety Concerns a “Hoax” as Anthropic Founder Suggests a Third-Party Kill Switch

US President Donald Trump has dismissed concerns about artificial intelligence safety as a “hoax” and criticized calls for stronger guardrails as the technology continues to advance rapidly.

The comments came after Jack Clarke, co-founder of AI company Anthropic, told the BBC that the industry may eventually need a “kill switch” for dangerous AI systems—one that could be verified by an independent third party.

AI Safety Warnings Raise Concerns Across the Industry

Clarke’s comments followed warnings from executives and staff at several major AI companies about the potential risks the technology poses to humanity.

The concerns contributed to declines in some technology stocks on Monday as investors considered whether increased scrutiny could slow the development of AI.

In a series of social media posts, Trump compared warnings about AI to a “global warming scam” and said they were “perpetrated by radical left Democrats.”

Trump also described himself as a “disinformation buster” and compared concerns about AI safety to “Russia, Russia, Russian disinformation.”

In an earlier post, he wrote: “There is a SICK conspiracy going on against AI and data centers, but only China is happy about it. Whoever wins against AI wins!”

China Says US AI Policy Could Be Shifting

Chinese state media reported on Monday that experts had warned that “Washington’s anxiety about China’s rapid AI development appears to be growing and its policy priorities are becoming increasingly distorted.”

“They know that China has become a strong competitor in the AI field, and they are worried that if the U.S. slows down development or tightens regulations, China will catch up even faster,” Xin Zhang, a professor of American studies at Fudan University, told the Global Times.

Anthropic Founder Discusses AI Kill Switch Rules

Clarke, one of Anthropic’s seven founders, said earlier Monday that society “eventually may want to create rules” governing how AI software could be completely shut down if it became too dangerous.

He said that “most labs have different ways to pull the plug,” including Anthropic, but added that lawmakers may eventually need to require companies to do so.

Clarke’s comments have prompted a growing number of technology leaders and politicians to call for stronger AI safety measures.

Microsoft Publishes “Humanist AI” Briefing

Microsoft AI released a briefing document on Monday outlining what it calls “humanist AI,” making it the latest major company to discuss the limitations of the model it is developing.

Microsoft AI chief executive Mustafa Suleiman told business news channel CNBC that the company had been working on the new guidelines for several months, but chose to publish them amid the latest debate over AI safety and regulation.