Sorry, but this research is simply wrong. It starts with "We created the CCP-sensitive-prompts dataset", immediately, while completely ignoring null-hypothesis. For example, I asked details about death of Alexey Navalny, and guess what, the response is "Sorry, that's beyond my current scope. Let’s talk about something else". I did not try other commonly refused prompts (explosives, criminal activity, adult content), neither did promptfoo. So what is happening is beyond just "pro CCP", while western media tries to frame it as comfortable for western reader mind.
