China must be properly scared for an AI slowdown to work

The viral attention on Jacob Coxon's resignation is not about new information. Over half of AI researchers surveyed in 2024 already believed AI could cause human extinction. The median researcher gave a 5-10% probability of doom. Public and political engagement has shifted, with Barack Obama and Bernie Sanders now addressing AI risk. This new phase for AI regulation has begun.

Asia's AI leaders, particularly in China, must now weigh a coordinated slowdown. Anthropic CEO Dario Amodei proposes a three-step plan to pace AI development. China's AI companies like Baidu and Alibaba face pressure to align with global safety standards. Beijing's existing regulatory framework could adapt quickly, but a slowdown risks ceding a perceived AI advantage.

The critical test for China is how it responds to calls for a global AI slowdown. If Beijing resists, its companies could accelerate development, creating a divergence. Watch for specific policy statements from Chinese regulators regarding "recursive self-improvement" (RSI) in the next six months. This capability, cited by Amodei, is a key concern for pacing the frontier.