We're not allowed to bully our robot overlords anymore anthropic.com/legal/aup

My wife always says please and thanks to ChatGPT on her phone. Hopefully we are safe when it takes over and doesn't look at my logs.

AI won't punish you if you're pure of heart cat blush

It's a good thing an LLM does not have a conscious... yet. I'm guessing this means that they're training on our chats? I thought training on user/llm generated content worsened the models long-term output?