Here is an investigation of how different queries are classified as hateful vs not hateful in ChatGPT: https://davidrozado.substack.com/p/openaicms

(2023)

	▲	fc417fc802 4 hours ago \| parent \| next [-]
		It's not due to a technological limitation but rather human imposed. Unless the social climate at OpenAI shifts it won't change.
	▲	Dig1t 5 hours ago \| parent \| prev \| next [-]
		Almost everything in this is still true with the latest models available today.
	▲	roysting an hour ago \| parent \| prev [-]
		[dead]