Claude was asked to create a site friendly for kids. But in a sort of paradox, it created a profanity filter that includes no real profanity, because it itself is not allowed to use the profanity.
Discussion
includes a it site use it kids. asked friendly But not is created profanity, profanity. no real of Claude sort in a allowed create that itself the because to for profanity filter was a to paradox,
I’ve gotten Claude to curse. It’s definitely capable, at least via API
This is Claude 3.7 —> ElevenLabs. The bleeps are where Claude cursed
What happens if you add the profanity yourself? Does it refuse to read the code 🤣
