The landscape of global artificial intelligence has undergone a seismic shift this week as prominent technology firms in China and Japan unveiled advanced AI models designed to compete directly with high-end American software currently restricted by export controls. Chinese cybersecurity giant 360 and Tokyo-based startup Sakana AI have both released frontier-level tools that aim to fill the vacuum left by Anthropic’s Mythos and Fable 5 models. These developments come just two weeks after the United States government, under the Trump administration, implemented a sweeping ban preventing non-Americans from accessing Anthropic’s most powerful cybersecurity-focused AI tools. The move by Asian firms signals a growing trend toward "AI sovereignty," as nations seek to insulate their critical infrastructure from the volatility of international trade policy and geopolitical tensions.
On Wednesday, the Beijing-based Qihoo 360 Technology Co. (360) introduced Tulongfeng, an artificial intelligence tool specifically engineered for high-level cybersecurity applications. According to reports, Tulongfeng is positioned as a direct competitor to Anthropic’s Mythos, a model so potent that it has been designated a national security concern by U.S. regulators. Alongside Tulongfeng, 360 also debuted Yitianzhen, a model built to automate cyber defense and incident response. The dual launch underscores China’s commitment to achieving self-sufficiency in the AI sector, particularly in domains that intersect with national security and software vulnerability detection.
Earlier in the same week, Sakana AI, a rapidly rising star in the Japanese tech ecosystem, launched Fugu. Named after the Japanese blowfish—a delicacy known for its complexity and potential lethality if mishandled—the Fugu model is marketed as a "frontier" AI that stands on equal footing with Anthropic’s restricted Fable 5 and Mythos Preview models. Unlike 360’s more aggressive positioning, Sakana AI has framed Fugu as an "orchestration model" designed to coordinate multiple AI agents and manage access to various models through APIs. This architectural choice is intended to provide Japanese businesses and government agencies with a robust alternative that remains immune to the sudden imposition of foreign export restrictions.
The Regulatory Catalyst: The US Export Ban on Anthropic
The catalyst for these rapid-fire releases was a mid-June 2026 order from the U.S. government that effectively severed global access to Anthropic’s premier models. Anthropic, a San Francisco-based AI safety and research company, had been on a historic trajectory, reporting a run-rate revenue of $47 billion in May 2026 and approaching a $1 trillion valuation ahead of a highly anticipated initial public offering. However, the capabilities of its Mythos model—specifically its proficiency in identifying and exploiting software vulnerabilities—led to concerns within the Trump administration regarding the potential for the technology to be used by foreign adversaries.
The resulting ban restricted Mythos and its slightly more limited counterpart, Fable 5, to American citizens only. This decision left enterprise customers across Asia, Europe, and the Middle East without access to what was considered the gold standard in cybersecurity AI. The ban has created an immediate market opportunity for local players who can offer similar capabilities without the "country risk" associated with U.S.-based providers. For many Asian firms, the sudden loss of access served as a wake-up call, highlighting the dangers of relying on a single foreign provider for critical technological infrastructure.
Sakana AI and the Strategy of Collective Intelligence
Sakana AI’s Fugu model represents a sophisticated response to this new reality. Founded in 2023 by former Google researchers David Ha and Llion Jones, along with former Mercari and Stability AI executive Ren Ito, Sakana AI has specialized in creating generative AI models optimized for the Japanese language and culture. The company recently raised $135 million in a Series B round, bringing its valuation to approximately $2.65 billion.
While Sakana AI’s spokesperson characterized the timing of Fugu’s release as "entirely coincidental," the company has not hesitated to leverage the current geopolitical climate in its marketing. Its website now explicitly advertises "frontier capability without the risk of export controls." David Ha, the company’s CEO, argued on social media that the era of relying on massive, centralized models is giving way to a more decentralized approach. "Orchestration models are the next frontier, beyond bigger models," Ha wrote on X. He emphasized that "access to top models can disappear overnight," and that "collective intelligence is the practical hedge against this concentration of power."
Fugu is designed to act as a conductor, managing various smaller, specialized models to achieve results comparable to larger, restricted systems. This approach allows users to maintain high-level capabilities while diversifying their technological dependencies. By targeting Japanese businesses and government agencies, Sakana AI is positioning itself as the primary guardian of Japan’s digital sovereignty.
Chinas 360 and the Doctrine of Strategic AI Assets
In China, the response to the U.S. ban has been even more pointed. 360’s founder, Zhou Hongyi, has been vocal about the necessity of domestic alternatives in the AI space. Upon the unveiling of Tulongfeng, Zhou described AI-driven vulnerability detection as a "national strategic asset." He warned against the risk of "one-way transparency," a scenario where the U.S. possesses advanced tools to find flaws in global software while denying those same tools to other nations, thereby creating a profound imbalance in global cybersecurity.
Tulongfeng is built to automatically discover software vulnerabilities at a speed and scale that exceeds human capability. By matching the performance of Anthropic’s Mythos, 360 aims to ensure that Chinese enterprises and government bodies remain on the cutting edge of cyber defense. The companion tool, Yitianzhen, further strengthens this position by automating the response to cyberattacks, reducing the time between detection and remediation. Unlike the "hedge" strategy employed by Sakana AI, 360’s approach is one of direct replacement and competitive parity.
Chronology of Events Leading to the Mid-2026 AI Shift
The emergence of Tulongfeng and Fugu is the culmination of a series of events throughout 2025 and early 2026 that have redefined the global AI market:
- Spring 2026: Sakana AI presents the underlying research for what would become Fugu at the International Conference on Learning Representations (ICLR).
- May 2026: Anthropic reports a record-breaking $47 billion run-rate revenue, signaling its dominance in the enterprise AI market.
- Early June 2026: The Trump administration issues an executive order citing national security concerns, banning the export of Anthropic’s Mythos and Fable 5 models to non-U.S. entities.
- Mid-June 2026: At the G7 summit in Evian, France, AI access and export controls become a central topic of discussion. Sakana AI co-founder Ren Ito participates in diplomatic talks regarding the preservation of technology access for allies.
- June 22, 2026: Sakana AI launches Fugu, marketing it as an export-control-free frontier model.
- June 24, 2026: 360 unveils Tulongfeng and Yitianzhen in China, explicitly positioning them as rivals to Anthropic’s restricted tools.
Analysis of Implications and Market Realignment
The launch of these models marks a potential permanent realignment in the global AI industry. While U.S. models have historically been the preferred choice for Asian enterprises due to their advanced capabilities, the reliability of that access is now in question. Ren Ito, in an op-ed for Project Syndicate, argued that the U.S. government’s first priority should be to preserve access for its closest allies. He warned that if AI becomes a "hoarded" technology rather than a collaborative one, it will force allies to develop their own independent ecosystems, potentially leading to a fragmented global market.
The economic stakes are particularly high for U.S. firms like Anthropic. While the company’s current revenue is massive, a significant portion of future growth was expected to come from Asian markets, particularly in tech-heavy hubs like Tokyo, Seoul, and Singapore. By creating a vacuum, the U.S. export ban has effectively subsidized the growth of competitors like Sakana AI and 360. These local alternatives offer a distinct advantage beyond just availability: they are often better trained on local languages, cultural nuances, and regional regulatory requirements.
Furthermore, the concept of "AI Sovereignty" is no longer a theoretical concern but a practical business requirement. For a multinational corporation or a national government, the risk of having a critical operational tool "turned off" due to a policy change in Washington is becoming unacceptable. This sentiment is driving investment into local AI labs across the globe, from the Middle East to Southeast Asia.
Conclusion and Future Outlook
As the "AI Cold War" intensifies, the rapid response from 360 and Sakana AI demonstrates that the technological gap between the U.S. and the rest of the world may be narrower than previously thought. While Anthropic’s Mythos may currently hold a slight edge in raw performance, the arrival of Tulongfeng and Fugu suggests that the monopoly on frontier-level AI is over.
The long-term impact of the current export ban remains to be seen. If the restrictions are lifted, U.S. companies may find it difficult to regain the trust of Asian customers who have already integrated local alternatives into their workflows. For now, the momentum has shifted toward regional players who can guarantee uptime and accessibility, regardless of the political climate in the United States. The era of global AI interdependence is facing its greatest challenge yet, replaced by a new era defined by strategic autonomy and localized innovation.
