I had Fable flag my question as unsafe with tag [bio] for asking it to ... use Unicode graphemes. In Elixir code that already was using this function elsewhere: https://elixir.hexdocs.pm/String.html
Happened to me with ChatGPT and Gemini too often, that a harmless change in my own picture is marked as harmful content (just changed the tie color). And the AI refuses and future edit.
