Why does your US-based AI response feel like Chinese censorship?

Let's start with a number: Over 2,500 companies. That's the scale of businesses affected by the same AI supply chain issue mentioned in that WSJ report. I stared at that figure for a while, and my first reaction wasn't shock—it was dread. When I wrote that piece titled 'Whose Code Is Your AI Running?' last week, I was still talking about how stealthy supply chain attacks are. Now this news gives me a much more concrete example: Chinese censorship has seeped into American AI responses through the pipeline of training data. It's not that the model learned bad habits on its own; it's that the data fed to it already came bundled with a value filter.