Sure, they might, but what you see in practice in GPT and being discussed in interviews by Sam is mostly the "AI shouldn't say uncomfortable things" version of

Sure, they might, but what you see in practice in GPT and being discussed in interviews by Sam is mostly the "AI shouldn't say uncomfortable things" version of AI "safety".