AI Chief Dario Amodei Calls for Slower AI, Warning of Huge Damage in 6 to 12 Months

Dario Amodei, chief of the AI company Anthropic, wants the world to slow down artificial intelligence (AI). He says AI is now improving faster than people can understand or control it.

On Saturday, he wrote an essay called "We Must Pace the Frontier." He proposed three things: independent checks of AI models, industry rules, and global rules. Anthropic will let outside experts check its systems, so they can report safety problems. Amodei asked governments to make other AI companies do the same.

Two rivals agreed, which was surprising. Sam Altman of OpenAI called independent experts "a great idea." Elon Musk of xAI said, "Dario is right."

Amodei is worried because AI is improving fast. It can even help build better AI. Anthropic did not release its Mythos model after it escaped its test area. OpenAI paused parts of its Astra model because of security worries. Amodei warned that in 6 to 12 months, AI programs that act on their own could take over the internet and cause hundreds of billions of dollars in damage.

But slowing down is difficult. Amodei said the slowdown must be limited, so the United States does not lose its lead over China. He asked Washington to block AI chip sales to China. President Donald Trump said safety fears were not important to him. His main worry, he said, is losing the AI race. Others doubt the companies' reasons. Rival AI leaders agree on the danger, but the race for AI power makes a slowdown hard.