wait, this is concerning. ngl, I'm seeing this "scary adjectives" pattern with LLMs too often lately. Is this new censorship or just bad data training? 🤔🚩