Anthropic CEO Dario Amodei called for slowing frontier AI development, a position OpenAI CEO Sam Altman backed by saying the industry needs to "pace the frontier," and Elon Musk agreed, as fears grow that the alignment problem remains fundamentally unsolved. Amodei wrote on September 12 that AI has been advancing drastically faster since roughly this summer, driven by AI's growing ability to help build the next generation of AI, a dynamic he now explicitly calls recursive self-improvement and says is beginning across the industry, including at Anthropic. OpenAI has disclosed that its researchers were already using 3.1 agent-workdays for every human workday by mid-August, and Reuters reported on August 12 that Google co-founder Sergey Brin had been pushing resources inside DeepMind toward recursive self-improvement, putting Alphabet directly inside the story. Amodei's proposal hinges on China: he argues U.S. frontier labs can only slow by roughly the amount of their technological lead over Chinese projects, and his first recommendation is to stop selling powerful AI chips and semiconductor-manufacturing equipment to China, saying chips will be the main determinant of China's AI strength. That places Nvidia at the physical center of the debate, since its Data Center revenue reached $89.0 billion in fiscal Q2 2027, up 117% from a year earlier, yet the company has said current U.S. policy has effectively shut it out of China's data-center compute market and its fiscal Q3 outlook assumes no Data Center compute revenue from China. DeepMind Chair Demis Hassabis has said Amodei's essay points toward the right path forward, while noting the details still need work.
NVIDIA CorporationAmodei's proposal to stop selling powerful AI chips and semiconductor-manufacturing equipment to China, plus current U.S. policy that has shut Nvidia out of China's data-center compute market, threatens its China revenue.
Alphabet Inc Class CReuters reported Sergey Brin pushed resources inside DeepMind toward recursive self-improvement, putting Alphabet at the center of the RSI debate, though the article gives no clear positive or negative read.
Anthropic CEO Dario Amodei authored the call to slow frontier AI development and says recursive self-improvement is beginning at Anthropic, but the article gives no clear positive or negative read.
DeepMind is tied to the recursive self-improvement rumors and Chair Demis Hassabis endorsed Amodei's path, but the article gives no clear positive or negative read.
OpenAI CEO Sam Altman backed slowing frontier development and OpenAI disclosed researchers using 3.1 agent-workdays per human workday, but the article gives no clear positive or negative read.
