Global AI Wars

The Five Eyes intelligence alliance issued a rare joint warning this week that frontier AI capable of crippling governments and businesses is close. The fast rise of Chinese and Japanese models helps explain the urgency and fear. Three new disruptions show just how fast it's happening:

1. It's becoming harder to put up a wall around America's advancements. Japan's Sakana AI launched Fugu Ultra, an orchestrator that it claims offers "frontier capability without the risk of export controls" by switching between publicly available models. The Tokyo-based company says it can reach Mythos-level performance by using U.S. labs' work as interchangeable infrastructure.
2. China is eating away at that lead by stealing America's best work. In February 2026, Anthropic accused DeepSeek, Minimax and Moonshot of illicitly training their own models via "distillation," using thousands of accounts to have millions of exchanges with Claude—a cheap shortcut to years of pricey research.
3. American labs are wondering whether the frontier is worth the risk. Two weeks ago, the Commerce Department export-controlled Anthropic's Fable 5 and Mythos 5, leading the company to shut off access for everyone. The best domestic AI companies may be wary to show off their highest-capability models, fearing further government intervention.