Asian AI firms launch Mythos rivals amid Anthropic export ban

Two major Asian AI companies released advanced models this week designed to compete with Anthropic's restricted technology, capitalizing on a U.S. export ban that has blocked international access to the company's most powerful AI systems.

Tokyo-based Sakana AI launched Fugu, a frontier model the company says matches the capabilities of Anthropic's Fable 5 and Mythos Preview. Days later, Chinese cybersecurity firm 360 Security unveiled Tulongfeng, an AI tool built for discovering software vulnerabilities that the company positions as a direct competitor to Mythos.

The releases follow a Trump administration order issued two weeks earlier that prevents Anthropic from providing global access to Mythos and its more restricted sibling, Fable 5. The ban applies specifically to non-American users, creating an immediate gap in markets that had relied on U.S. AI infrastructure.

Sakana positions model as hedge, not replacement

A Sakana spokesperson told TechCrunch the timing was "entirely coincidental," though the company's website now advertises "delivering frontier capability without the risk of export controls."

The model represents more than opportunism, according to Sakana CEO David Ha. Fugu is designed as an orchestration system that coordinates agent usage across multiple models through their APIs, rather than functioning as a standalone replacement.

"Access to top models can disappear overnight," Ha wrote, describing the approach as "the practical hedge against this concentration of power."

Sakana co-founder Ren Ito, speaking at last week's G7 summit in Evian where AI export controls dominated discussions, emphasized that U.S. models "remain important to Asia." In a Project Syndicate op-ed, he urged American policymakers to "preserve access" for close allies rather than treating AI as "a technology that is hoarded."

Founded in 2023 by former Google researchers including Ito, Llion Jones, and David Ha, Sakana builds generative AI models optimized for Japanese language and culture that work efficiently with smaller datasets. The company is targeting Japanese businesses and government agencies seeking to reduce exposure to tightening export restrictions.

China's 360 takes harder line

China's 360 Security adopted a more assertive stance. According to Reuters, founder Zhou Hongyi described vulnerability-detection AI as a national strategic asset and warned of "one-way transparency" risks where some nations possess advanced security capabilities while others remain exposed.

Alongside Tulongfeng, 360 released Yitianzhen, a model designed to automate cyber defense and incident response. The company did not respond to requests for comment.

Why it matters

The rapid emergence of competitive alternatives reveals how quickly export controls can accelerate rather than contain AI development in restricted markets. Anthropic reported $47 billion in annualized revenue as of May 2026, though the company has not disclosed what portion depends on Asian enterprise customers. Even if U.S. restrictions are eventually lifted, locally developed models trained on regional languages and cultural contexts may prove difficult to displace. The episode demonstrates that AI access restrictions create immediate market opportunities for competitors in affected regions, potentially undermining the strategic goals such controls aim to achieve.

The developments were first reported by TechCrunch.