Why the Real AI Race Just Became a Race to Lock Down Code, Not Build It
The thing that made America's lead in AI feel safe was never just having the best models — it was keeping the best ones out of rivals' hands. That advantage is now narrowing on two fronts at once: China has reportedly closed the cybersecurity capability gap with Anthropic, the leading U.S. AI lab, just as Washington restricts who can even use Anthropic's newest model.
Bottom Line
The story isn't 'China caught up' — capability gaps always close. It's that America's edge depended on controlling access, and the same restriction meant to protect that edge may now mostly disadvantage U.S. defenders without slowing Beijing. When both sides can build the offense, the contest shifts to who can deploy the defense fastest and widest — and a narrow, gated release works against that.