There's no single Chinese censorship program. Every company has their own process for deleting content, and they don't always make the same decisions. At least

There's no single Chinese censorship program. Every company has their own process for deleting content, and they don't always make the same decisions. At least the biggest platforms definitely use a lot of automation. E.g. I follow Professor Hu Fanzhu of East China Normal University's National Discourse Ecology Research Center on WeChat. A week ago, he published an article "Nine questions about Zhengzhou" that appeared in my feed, but displayed the message that it couldn't be sent because it might be inappropriately using the names of government institutions or employees. This was probably automated. Then he republished it as "Nine questions about ZZ", which displays the message that it was taken down after being reported by users. He's now switched to publishing a series of articles (five so far, here's the first: https://mp.weixin.qq.com/s/Rf2t5cRoHJGh8Yb-3fFXSA https://mp.weixin.qq.com/s/Rf2t5cRoHJGh8Yb-3fFXSA ) of which the second was taken down, again mentioning user reports as the reason. But I actually managed to read that one before it disappeared. To generalize a bit, any new event escapes censorship at first, because neither automated systems nor human reviewers have been instructed to censor it. Once the censors are aware, new keywords like "Zhengzhou" can be added to the filter lists. But those are easy to get around by using different words, so humans are still needed to clean up the rest. And it can take them a long time to do so, during which tens of thousands of people may have read something the government didn't want them to know about.