July 20, 2026
OpenAI? More like Open Fight
American AI is locked down and proprietary. It's losing
China’s free AI play has commenters fighting over hype, trust, and who’s really trapped
TLDR: The article argues China is gaining ground in AI by releasing cheaper, more portable models while American companies keep theirs locked behind closed doors. Commenters are split between calling it a smart winning strategy, a suspicious giveaway campaign, or just more proof that nobody trusts any AI to stay neutral.
The big claim from this Verge report is simple: Chinese AI companies are giving away more portable models while American giants keep their best tools locked inside paid platforms, and that could backfire badly. The argument is that if AI becomes easy to swap—today one chatbot, tomorrow another—the real money won’t be in the model itself, but in the business services built around it. And that’s where commenters smelled both opportunity and chaos.
The hottest reaction? A full-blown trust war. One side says China’s strategy looks like a classic “give it away cheap, crush the competition later” move. Another side fired back that American AI isn’t exactly neutral either, mocking the idea that only Chinese models come with political baggage. In other words: everyone’s accusing everyone else of bias, and nobody’s pretending these tools are just harmless little robot helpers anymore.
Then came the skepticism brigade. One commenter flat-out side-eyed the claim that “80% of startups” are using Chinese models, saying the startups they interviewed with were still living in Claude-and-OpenAI land. Others raised the security panic button: even if a Chinese model is run on American soil, could it still generate risky code or leak information if extra features are turned on?
And yes, the thread delivered comedy too. The standout joke-test was asking a model to write a tiny program called FreeTaiwan() and declaring that if it refuses, “it fails. End of story.” Dark, petty, extremely online—and basically the perfect summary of the mood.
Key Points
- •The article argues that AI models themselves have limited moat because users can switch between vendors relatively easily, especially through APIs.
- •It says enterprise value is more likely to come from surrounding services such as contracts, integrations, and workflow features rather than the model alone.
- •The article states that U.S. export controls on GPUs and data-sharing restrictions have pushed Chinese firms toward open-weights distribution instead of centralized global services.
- •It claims open-weights models provide portability and permissionless use, creating broader ecosystem and distribution advantages.
- •The article cites recent releases from Moonshot and Alibaba, along with comments from Martin Casado, as signs that Chinese models are closing the gap with leading U.S. systems.