Anthropic outlines how the US could gain a 1-2 year AI lead on China (www.businessinsider.com)

Anthropic has issued a critical call for the US to swiftly secure a 1-2 year lead in AI development over China, emphasizing the importance of tightening policies on chip exports. The company analyzed two potential scenarios for the AI landscape in 2028: one in which the US restricts China's access to American AI computational power and another where it does not. They argue that China is narrowing the AI gap through lenient chip export controls and distillation attacks—methods that allow them to use existing advanced AI models to train smaller, derivative models. This situation highlights the urgency for the US to tighten export controls and enhance enforcement mechanisms to prevent technology leakage, as a competitive race could compromise safety protocols in AI development. Anthropic warns that without these measures, the US risks losing its edge, making it easier for the Chinese Communist Party to catch up in advanced AI capabilities. The discussion comes amid ongoing tensions and evolving regulatory landscapes, underscoring that maintaining a lead in frontier AI is not just about technology but also governance and safety protocols.