Chinese State Media Accuses Anthropic of AI Double Standards

Chinese state media has leveled accusations of double standards against US AI lab Anthropic, as planned US-China talks on AI governance could impact model access, monitoring, and enterprise deployment. A CCTV-affiliated social media account alleged US AI firms apply different rules to Anthropic, singling it out for criticism. This comes amid reports of potential AI governance discussions ahead of Xi Jinping's upcoming visit to the US on September 24.

The Chinese commentary accused Anthropic of attempting to set global AI rules through lobbying and warning about militarized AI development, calling for the US government to differentiate between genuine security threats and legitimate tech competition. Anthropic has been vocal in accusing Chinese AI model developers of illicit "model distillation," a process where developers use thousands of bot accounts to generate massive data exchanges with larger models and then use those responses to train their own models.

While Anthropic's confrontational stance stands out, industry figures argue for continued access and collaboration. OpenAI’s head of strategic futures acknowledged the ground has shifted, suggesting a window for US-China AI safety cooperation. The growing popularity of Chinese AI models has heightened policy scrutiny in both nations, with US officials warning of sanctions against Chinese AI labs using frontier US models.

The talks may help ease tensions and potentially allow both countries to reach a form of understanding, though an agreement may not favor US companies as much due to China's leverage through popular AI models. The outcome could affect enterprise AI teams, determining cross-border availability of models, monitoring systems, and potential collaborations between Chinese open-weight models and US commercial platforms.