China Hits Back at US AI Lab Anthropic as Tensions Rise

Chinese state media has accused Anthropic — the American company that makes the Claude AI assistant — of playing by different rules depending on the country. The criticism came through a social media account linked to the state broadcaster CCTV. It said Anthropic hides a "desire for control" behind a "saviour complex" and tries to shape global AI rules while pushing its own interests. This is happening just before a planned September visit by China's president to the US, where AI governance could be on the table.

At the heart of the dispute is a practice called "model distillation." That's when a developer uses thousands of fake accounts to ask a powerful AI millions of questions, then uses those answers to train a cheaper rival AI. Anthropic says Chinese companies like Alibaba and DeepSeek are doing this on an industrial scale, stealing its work. But China's media points out that distillation is common even among American firms — so why are Chinese companies being singled out?

Meanwhile, not everyone in the US tech world wants a fight. OpenAI, usually a tough competitor to Anthropic, has said some Chinese AI models are impressive and argued for continued access. Industry groups want fewer restrictions on open AI models, which anyone can use and modify. The reason is simple for businesses: some Chinese models are good enough for specific tasks and cost less than one-tenth of the price of leading American models. That price gap matters for startups and companies watching every dollar.

The talks could lead to new rules on which models are available in each country — similar to earlier disputes over computer chips, where the US blocked the most advanced chips from going to China. For everyday users and companies, the outcome will affect the AI tools they can choose and how much they pay. If tensions rise, some cheap or powerful models could disappear from the market. If talks go well, you might see more options and lower prices.