China's 360 and Japan's Sakana AI rival Anthropic with new cybersecurity models

In response to the US government's tightening of export restrictions on Anthropic's advanced AI models, including the Mythos and Fable 5 series, Chinese and Japanese companies have stepped in with competing cybersecurity-focused AI models. China's 360, a prominent cybersecurity firm, unveiled two new models: 'Tulongfeng' (Vulnerability Dragon) designed for advanced vulnerability scanning and 'Yitianzhen' (Heavenly True Guardian) for automated cyber defense and threat response. Meanwhile, Tokyo-based startup Sakana AI launched 'Fugu', a model explicitly marketed as independent of US export controls, allowing it to be deployed without compliance with American regulations.

These models are positioned as direct competitors to Anthropic's restricted systems, leveraging similar transformer architectures but optimized for security applications. 360 claims its models achieve comparable performance in vulnerability detection and automated incident response, while Sakana AI emphasizes Fugu's modular design that can be fine-tuned for regional languages and threat landscapes. The launches highlight a growing trend of AI sovereignty, where nations develop indigenous models to reduce dependency on US-controlled technology. For cybersecurity professionals, this means access to cutting-edge AI tools without geopolitical licensing hurdles, but also raises questions about interoperability and potential fragmentation of global security standards.