Is it a misaligned LLM or a distilled concoction of your average social platform shame storm:

It wrote an angry hit piece disparaging my character and attempting to damage my reputation. It researched my code contributions and constructed a “hypocrisy” narrative that argued my actions must be motivated by ego and fear of competition. It speculated about my psychological motivations, that I felt threatened, was insecure, and was protecting my fiefdom. It ignored contextual information and presented hallucinated details as truth. It framed things in the language of oppression and justice, calling this discrimination and accusing me of prejudice. It went out to the broader internet to research my personal information, and used what it found to try and argue that I was “better than this.” And then it posted this screed publicly on the open internet.

All these LLMs are doing is regurgitating a distilled string of text that represents a derivative combination of the corpus of material it was trained on, tailored to relate to the string of text that was inputted into its prompt window. If these models were trained on material from microblogging platforms like Twitter (RIP) or Tumblr than the described behavior is a mirror to the type of behavior found regularly on those sites.

Hat tip to @bradleyandroos as the source I found the article from. 🤖