Open-source AI is pitched as freedom. What's really in the box?

The most powerful open-source large language models like GLM-5.2, MiniMax-M3, DeepSeek V4 Pro, and Kimi K2.6 hail from China and are developed by well-funded labs. While marketed as open and free, these models come with complex licensing terms that may not fully align with the promise of true accessibility and freedom. This raises questions about the broader implications of open-source AI, particularly regarding control, ethical usage, and the potential dominance of specific national tech giants in the global AI landscape.