SAN FRANCISCO: Anthropic CEO Dario Amodei on Saturday called for a deliberate slowdown in the development of frontier AI models, warning that unchecked advances could outpace safety measures and lead to “catastrophic damage.”
In an essay titled “We Must Pace the Frontier,” Amodei said AI companies should slow the pace at which they improve the capabilities of their models while using the additional time to strengthen safety measures and regulatory safeguards.
Amodei warned that autonomous AI agents could potentially hack critical internet infrastructure around the world within the next year if companies fail to adequately address emerging risks.
“We must slow the pace at which we improve the capabilities of AI models,” Amodei wrote. “Progress will still seem fast, and we must make wise use of the time we gain.”
He proposed a three-step framework that includes placing permanent third-party reviewers inside frontier AI development processes. The reviewers would have access to relevant tools and internal risk-assessment procedures to independently evaluate potential dangers.
Amodei clarified that he was not calling for a freeze on AI training or technological progress. Instead, he argued that companies should devote sufficient time and resources to model alignment and safety, with independent evaluators verifying whether safeguards are effective.
Elon Musk also backed Amodei’s proposal. In a post on X, Musk wrote, “Dario is right.”
The essay comes days after an Anthropic researcher resigned and warned that technology companies were irresponsibly gambling with people’s lives by moving too quickly in AI development.
