Sakana AI, 360 Security, Z.ai Ship Direct Alternativas a los modelos antropogénicos prohibidos

Asian startups — Sakana AI's Fugu, 360 Security's Tulongfeng, and Z.ai's updated GLM models — are now directly competing with Anthropic's restricted Mythos and Fable 5 in enterprise cybersecurity, less than a month after US export controls kicked in.

US restrictions on Anthropic's most capable cybersecurity AI models, driven by national security concerns, were supposed to keep sensitive capabilities out of certain overseas markets. Instead, they lit a fire under domestic AI development in Japan and China. Companies need vulnerability detection, malware analysis, and security research tools — they aren't waiting for Washington to sort out license approvals.

Fugu, Tulongfeng, and GLM: Three Asian Alternatives

Tokyo-based Sakana AI launched Fugu, an orchestration-based cybersecurity model that the company says delivers performance comparable to Anthropic's Fable 5 on several industry benchmarks. Rather than training a single frontier-scale system, Sakana orchestrates multiple models — a lower-cost path to competitive results.

In China, 360 Security released Tulongfeng, an AI-powered vulnerability discovery model designed to go head-to-head with Anthropic's Mythos. Chinese AI startup Z.ai also demonstrated cybersecurity improvements in its latest GLM models. All three are targeting the gap left by restricted access to Anthropic's top-tier systems.

Export Controls Reshape the AI Map

This mirrors the semiconductor playbook: when you cut off access, you force domestic investment. Asian startups aren't just competing on cost anymore — they're building specialized models for enterprise security use cases. Governments now treat advanced AI as strategic technology, and export controls are accelerating the decentralization of innovation.

Anthropic has reportedly restored limited access to some trusted organizations, but broader restrictions remain in place. That window is enough for Asian firms to establish themselves in a market that was previously dominated by a handful of US companies. If you're a CISO evaluating cybersecurity AI tools, you've now got three new names on your shortlist — and they're not waiting for export licenses.

Source: Anthropic's export restrictions are fueling Asia's AI boom
Domain: economictimes.indiatimes.com