OpenAI's limited release of GPT 5.6 highlights the Trump administration's push for AI regulation, with CEO Sam Altman acknowledging the process falls short of optimal. The rollout reflects broader tensions between Washington's oversight ambitions and the rapid pace of frontier model development.

Chinese competitors are advancing quickly. Z.ai's GLM 5.2 model impressed David Sacks, co-chair of Trump's science advisory council, who noted its capabilities rival current offerings from leading US labs. While US companies maintain advantages in complex, long-horizon tasks, the gap is narrowing.

Traditional arms control frameworks face fundamental obstacles. Cyber operations already widely deploy AI models from both nations, making restrictions impractical. Biosecurity represents another potential focus area, though China's transparency record raises concerns about verification.

The "trust but verify" challenge proves even more acute with AI than nuclear weapons. Without meaningful compliance mechanisms, both Washington and Beijing will prioritize competitive advantage over cooperative safety measures.