Bloomberg: China is concerned that cutting-edge models from the U.S. like Anthropic may be used as tools for cyber attacks against China

According to Bloomberg, citing informed sources, Chinese officials are concerned that cutting-edge AI models from the U.S., such as Anthropic's Mythos, could be used as offensive cyber tools to launch cyberattacks against China. At the same time, the Chinese side questions why Anthropic does not allow Chinese companies to use the relevant models normally, yet cannot rule out the possibility of these capabilities being used to attack China. Anthropic has long advocated for the U.S. to tighten restrictions on advanced chips and AI capabilities related to China, and last year it stopped providing services to companies controlled by China.

The report states that the Chinese side is currently not prepared to take immediate action, but if the U.S. further restricts Chinese AI companies, relevant departments are studying response plans, which may include imposing sanctions on U.S. AI companies or placing them on a restricted list. Currently, neither Anthropic nor relevant departments have publicly responded to this.