What's interesting to me is #deepseek is censoring some queries at the prompt level, and some it lets it get through the reasoning, then censors after it's visualized the reasoning and answer, which is a very funny form of censorship?

I don't have much to contribute on here, but here's some Billy content