China Takes Note of Anthropic's Mythos Model After Initial Skepticism

In the third of a three-part series on Anthropic's powerful Mythos AI model, we look at how China is responding to the growing threat of AI-powered cyberattacks and what it means for the country's cybersecurity. China is not immune to any fallout that might result from AI-powered cyberattacks, according to experts, although a sense of panic may not be as high due to the country's strict scrutiny of its artificial intelligence industry. In the weeks since US start-up Anthropic announced its new AI model, Claude Mythos Preview, China has largely been an observer, even as foreign governments and regulators have rushed to warn industries and companies about the potential implications. Instead, many in China accused the company of hype by citing unprecedented cybersecurity risks as justification for limiting model access to a small group of US tech companies and the US government. On Zhihu, China's most popular Q&A platform and a discussion board for AI enthusiasts, many users dismissed Anthropic's warnings in discussions that garnered nearly 2 million views, pointing to past instances of US AI companies using fear as a marketing ploy. How has China's perception of Mythos shifted? However, recent developments have led some in China to take the model more seriously. On Sunday, Beijing commented on Mythos directly for the first time in an article on Yuyuan Tantian, an influential social media account run by state broadcaster CCTV, saying that the model showed "unprecedented cyberattack capabilities".