The global artificial intelligence landscape is undergoing a profound geopolitical realignment, catalyzed by strict regulatory measures and tightening trade restrictions imposed by the United States. In the wake of the Trump administration’s sudden decision to ban the global export of Anthropic’s cutting-edge cybersecurity AI model, Mythos, and its restricted counterpart, Fable 5, international competitors have swiftly moved to fill the void. Within days of the U.S. directive, major technology entities in both Japan and China unveiled competing frontier models explicitly engineered to bypass American export controls while addressing regional technological sovereignty.
On Wednesday, Chinese cybersecurity heavyweight 360 introduced Tulongfeng, an advanced artificial intelligence tool designed to rival Anthropic’s restricted capabilities. This rollout followed closely on the heels of a Tokyo-based AI startup, Sakana AI, which launched its own frontier model, Fugu, earlier in the same week. Together, these developments highlight a growing urgency among Asian economies to reduce their technological reliance on U.S. infrastructure, a vulnerability exposed dramatically by sudden regulatory shifts in Washington.
The Catalyst: The U.S. Export Ban on Mythos and Fable 5
The geopolitical friction driving these regional developments stems from a policy decision enacted roughly two weeks prior, when the U.S. government restricted international access to Anthropic’s elite cybersecurity models, Mythos Preview and Fable 5. American authorities cited critical national security concerns regarding the immense autonomous capabilities of these models, particularly their dual-use potential in offensive and defensive cyber operations.
Anthropic, which experienced meteoric commercial growth—reporting a run-rate revenue surpassing $47 billion by May 2026 and nearing a historic $1 trillion valuation ahead of an anticipated IPO—found its global expansion abruptly checked. While the financial impact of losing access to key enterprise markets in Asia remains to be fully quantified, the regulatory vacuum created by the decision was instantaneous. Rather than halting their technological advancement, the restriction galvanized international competitors to accelerate local alternatives.
Sakana AI and the Rise of Collaborative Agent Ecosystems
Sakana AI, a Tokyo-based startup co-founded in 2023 by former Google researchers David Ha and Llion Jones alongside former Mercari and Stability AI executive Ren Ito, launched its Fugu model to immediate international attention. Named after the Japanese word for blowfish, Fugu is designed as an agentic frontier model capable of orchestrating access to multiple underlying AI models via Application Programming Interfaces (APIs).
Despite the fortuitous timing of the release—coming immediately after the U.S. restriction on Anthropic—Sakana representatives maintained that the launch was entirely coincidental. A company spokesperson noted that the foundational research behind Fugu had been presented at the International Conference on Learning Representations (ICLR) earlier in the spring, culminating from months of systematic development since late last year.
Nevertheless, Sakana has leaned into the market reality. The company’s official promotional materials emphasize its ability to deliver "frontier capability without the risk of export controls." By focusing on generative AI models that are computationally efficient, effective on smaller datasets, and culturally optimized for Japanese language use, Sakana is positioning itself as a reliable partner for Japanese businesses and government agencies seeking operational continuity.
However, Sakana’s leadership has been careful to frame Fugu not as a permanent, adversarial pivot away from Western technology, but rather as a strategic hedge against supply chain volatility. Co-founder and CEO David Ha emphasized on social media that the future of enterprise AI lies in orchestration rather than monolithic reliance on a single provider.
"Orchestration models are the next frontier, beyond bigger models," Ha stated, pointing out that relying on a single geographic jurisdiction for national digital infrastructure poses an unacceptable risk. "Access to top models can disappear overnight. Collective intelligence is the practical hedge against this concentration of power."
This nuanced diplomatic stance was echoed by co-founder Ren Ito during the G7 summit in Evian, where cross-border data governance and export controls dominated international discussions. In a subsequent op-ed published via Project Syndicate, Ito urged Western policymakers to recognize that technology sovereignty is fundamentally about preservation of options rather than total ownership, arguing that artificial intelligence should be collaboratively developed rather than hoarded.
China’s 360 Enters the Fray with Strategic Cybersecurity Tools
While Japanese startups are framing their new models as collaborative hedging strategies, Chinese firms are taking a more direct and assertive approach to national technological self-reliance. On Wednesday, the Chinese cybersecurity titan 360 announced the development of Tulongfeng, positioned explicitly to compete with Anthropic’s Mythos. Alongside Tulongfeng, 360 introduced Yitianzhen, a specialized automation tool engineered for comprehensive cyber defense and rapid incident response.
Unlike general-purpose generative models, these tools target the highly sensitive domain of automated software vulnerability discovery and threat mitigation. According to reports from Reuters, 360 founder Zhou Hongyi framed vulnerability-discovering artificial intelligence as an essential national strategic asset. Zhou warned against the dangers of "one-way transparency"—a scenario where select geopolitical actors retain exclusive access to advanced offensive and defensive cyber capabilities while leaving other nations exposed.
Unlike Sakana AI, which has maintained open channels with Western stakeholders, 360 declined to comment on its latest releases, reflecting the broader tech decoupling occurring between Beijing and Washington.
Chronology of Events
- Spring 2026: Sakana AI presents the foundational research behind its frontier agent architecture at the International Conference on Learning Representations (ICLR).
- May 2026: Anthropic announces historic financial milestones, crossing a $47 billion run-rate revenue threshold ahead of its planned public offering.
- Early June 2026: The U.S. federal government enacts a strict ban prohibiting the export or non-American access to Anthropic’s advanced models, Mythos and Fable 5.
- Mid-June 2026: G7 leaders convene in Evian, where Sakana co-founder Ren Ito addresses international delegates on the risks of AI hoarding and the importance of open access for allied nations.
- Late June 2026: Sakana AI officially launches Fugu to global attention. Days later, Chinese cybersecurity firm 360 unveils its competing suite of vulnerability-assessment AI tools, Tulongfeng and Yitianzhen.
Market Implications and the Long-Term Outlook
The simultaneous emergence of Fugu in Tokyo and Tulongfeng in Beijing signals a permanent shift in the global artificial intelligence economy. The era of unhindered, U.S.-led technological hegemony over frontier AI models is facing structural friction from sovereign governments determined to insulate their critical infrastructure from geopolitical whims.
Analysts suggest that even if the United States were to lift its export restrictions on models like Mythos and Fable in the future, the trust deficit has already been established. Foreign enterprises and public sector agencies are increasingly unwilling to build critical workflows on platforms vulnerable to overnight revocation. By investing heavily in localized training data, regional language optimization, and multi-model orchestration frameworks, Asian technology firms are rapidly establishing a resilient, parallel ecosystem.
Ultimately, the U.S. export controls intended to secure technological supremacy may have inadvertently accelerated the fragmentation of the global AI market, ensuring that the future of artificial intelligence will be multipolar, competitive, and fiercely regionalized.
