Anthropic's Claude will end harmful chats like a boss

Here is the pr from anthropic: https://www.anthropic.com/research/end-subset-conversations

The company says it’s implementing this protection not primarily for users, but for the AI model itself, exploring the emerging concept of “model welfare.”

This just sounds like blatant marketing to be honest.