So the thing to note about Tiananmen square outputs is bias.
Our datasets have bias. The way we program has the bias of our backgrounds and politics.
It's why the OSI open AI definition isn't worth the paper it's written on.
You need to examine what data it's been trained on.
But well done Guardian news doing a video comparison of deepseek, ChatGPT and Gemini. *Slow clap*
I'll await more breathless reporting from you, as you fail to examine deeper.