Experts Raise Concerns Over China's SOTA AI Models, Including QwQ and Deepseek-r1

Concerns are growing over the rapid advancements in China's open-source AI models, such as QwQ and Deepseek-r1, which some experts claim are surpassing those developed by companies like Anthropic and Google. These models are considered state-of-the-art (SOTA). Aza Raskin and Tristan Harris have pointed out that Meta's open-source AI model has significantly accelerated China's AI development, while OpenAI's models have been criticized for insufficient security measures against potential theft. HuggingFace CEO Clement Delangue has expressed reservations about building on China's open-source AI models, citing issues such as how these models might respond differently to politically sensitive topics like Tiananmen Square. These developments highlight ongoing tensions in the global AI landscape, with calls for more robust security and regulation in the sector.