Asian AI labs rush in as US export ban on Anthropic's Mythos drags on

Two weeks after a Trump Administration order barred Anthropic from giving non-Americans access to its cybersecurity-focused Mythos model and the more restricted Fable 5, rival labs in Japan and China have moved to fill the vacuum. Tokyo-based Sakana AI launched Fugu, a frontier model it claims rivals Fable 5 and Mythos Preview, pitching it to Japanese businesses and government agencies as a way to sidestep tightening export controls. China’s 360 went further, unveiling Tulongfeng, a tool for automated vulnerability discovery it positions against Mythos, alongside Yitianzhen for automated cyber defense and incident response.

The two launches reflect sharply different postures. Sakana frames Fugu as a hedge rather than a replacement, with CEO David Ha arguing that depending on a single provider for national infrastructure is an unacceptable risk and that orchestration models coordinating many AIs are the next frontier. Co-founder Ren Ito, echoing remarks at the G7 summit, has publicly urged Washington to preserve allies’ access and treat AI as something developed collaboratively rather than hoarded. 360, by contrast, isn’t hedging: founder Zhou Hongyi called vulnerability-finding AI a national strategic asset and warned of “one-way transparency,” where some actors gain advanced offensive-detection capabilities that others lack.

The stakes are commercial as well as geopolitical. Anthropic reported run-rate revenue crossing $47 billion in May 2026, though its exposure to Asian enterprise customers is undisclosed. The broader risk for US labs is durability: even if the ban is eventually lifted, locally trained alternatives tuned to regional language and nuance are already establishing footholds, and trust lost during the freeze may not return cleanly.