Predictable: when chatbots censor speech Big Data disagrees with, it’s called ‘guardrails’; when it censors speech they agrees with, ooh, then very bad…
No, it couldn't. You've got to learn what LLM "alignment" is, the frontier labs are heavily invested in it.