The clash sharpened on Sept. 9, when former Anthropic researcher Jacob Coxon wrote that people building AI “earnestly believe that it could kill us all by the end of the decade.” Anthropic alignment-science lead Evan Hubinger backed that assessment, saying he personally put the chance of AI killing all humans at more than 10% within a decade.¹

Five days later, President Donald Trump framed restraint as a geopolitical risk, arguing that the U.S. already had ample criminal and regulatory authority over big technology companies and warning that opposition to AI and data centres benefited China.¹

Then, in a Sept. 17 CBS interview, Huang delivered the bluntest possible rebuttal: “There is zero percent chance that’s going to be the end of the world.” He called scaring people “unnecessary” and “irresponsible,” and said existential-risk predictions were not grounded in science.¹ Yann LeCun amplified a post quoting Huang’s “0% chance” assessment and suggesting that AI “fearmongering” might be politically motivated or attention-seeking.Yann LeCun (ylecun)²

Huang’s policy prescription matched Trump’s accelerationist posture. Rather than create AI-specific rules, he argued, governments should enforce existing cybersecurity, product-liability and unauthorized-access laws first.¹

In a later interview with Ezra Klein, Huang widened his attack on the doomer camp, targeting Geoffrey Hinton’s estimate of a 10% to 20% risk of societal collapse. “That 10 percent chance is not grounded on science. It’s not grounded on research,” Huang said, adding that an alarmist track record was “literally horrible.”³

The divide is not over whether AI demands responsibility; it is over what uncertainty requires. Huang says evidence should restrain alarm. His critics say the absence of reliable probabilities is precisely why a potentially irreversible threat cannot be waved away.¹