> Of course, yes, the models will provide propaganda-aligned responses to prompts that specifically mention certain political issues. The only study I’ve seen

> Of course, yes, the models will provide propaganda-aligned responses to prompts that specifically mention certain political issues. The only study I’ve seen about that actually showed Chinese models to be less “propaganda-aligned” than the Western counterparts.