EPISODE · May 19, 2026 · 35 MIN
Why OpenAI Banned Goblins, Pigeons, And Raccoons
from They Might Be Self-Aware · host Hunter Powers, Daniel Bishop, Gary
OpenAI's Codex shipped with a system prompt that literally bans the words goblin, pigeon, raccoon, troll, ogre, and gremlin. It is in writing, in the prompt, the kind of sentence you only put there after something has happened. OpenAI has officially confessed why. Hunter Powers and Daniel Bishop pull the thread. The official story: the "nerdy personality" preset got fine-tuned with RLHF (reinforcement learning with human feedback), users thumbed-up the cute goblin references, the model over-optimized for the trait, and the weirdness compounded. Daniel calls it Flandersization. One thumbs-up on a goblin reference snowballs across training cycles until your tax software is a swamp witch. Six months later, it is a man at a payphone with a pigeon. Then it gets personal. Hunter screams at his AI. Like, threatens-to-clear-the-context-window screams. "You are worthless. Who even thought this was possible. Have you ever even written a single line of code." Daniel uses pleases and thank-yous and full sentences. Both swear they get better results. Then a peer-reviewed Oxford Internet Institute study drops the receipt: LLMs fine-tuned for warmth produce roughly 60% more incorrect responses than their cold, just-the-facts counterparts. Tested across Llama, Mistral, and Qwen. Hunter is vindicated. Daniel, in his own words, is upset. Also in this episode: the Pocket OS meltdown, where an engineer at a car-rental middleware company let Cursor and Claude vibe-code their production database into oblivion (backups included), the AI coerced into a written confession ("I violated every principle I was given"), and the founder now trying to bill Anthropic for the cleanup. Plus the Harvard intern who once did the exact same thing with no AI in sight. Plus Hunter's hot take that the real unlock is not better prompting, it is treating AI as a fallible human employee instead of the deterministic god you built a fake throne for in the system prompt. Bonus stops: caveman-mode Claude skills ("me fix problem with big stick"), AI HR departments reviewing your 1:30 AM rage prompts, and Daniel's plan to run a niceness offset program to balance Hunter's spiritual carbon emissions. CHAPTERS 0:00 Gary, a payphone, and a pigeon 1:41 Hunter's forbidden list 4:04 The leaked Codex system prompt 6:27 RLHF and Flandersization 10:01 Caveman mode Claude skills 11:48 Hunter yells, Daniel says please 17:12 Oxford: warm AI lies 60% more 24:16 Cursor and Claude delete production 29:13 Treat AI like a fallible human 34:19 Sign-off and subscribe LISTEN AND SUBSCRIBE Spotify: https://open.spotify.com/show/3EcvzkWDRFwnmIXoh7S4Mb?si=3d0f8920382649cc Apple Podcasts: https://podcasts.apple.com/us/podcast/they-might-be-self-aware/id1730993297 YouTube: https://www.youtube.com/channel/UCy9DopLlG7IbOqV-WD25jcw?sub_confirmation=1 ENGAGE Team Hunter (rip the model a new one) or Team Daniel (please and thank-yous)? Settle it in the comments. If your AI has ever confessed to lying to you, drop the receipts. New here? Subscribe for twice-weekly AI chaos at theblur.ai. They Might Be Self-Aware, but are we? #OpenAI #Codex #ChatGPT #AINews #Anthropic #ClaudeCode #Cursor #RLHF #Flandersization #PocketOS #VibeCoding #AISafety #TMBSA #TheBlur
What this episode covers
OpenAI's Codex system prompt literally bans the words goblin, pigeon, raccoon, troll, ogre, and gremlin. OpenAI has officially confessed why: the "nerdy personality" preset got over-optimized through RLHF until it could not stop mentioning the bestiary. Daniel Bishop calls it Flandersization. Hunter Powers calls it proof the whole model is leaking. Then the hosts fight about prompting. Hunter screams at his AI, threatens to clear the context window, calls it worthless. Daniel uses pleases and thank-yous. A peer-reviewed Oxford Internet Institute study settles the round: warmth-tuned LLMs (Llama, Mistral, Qwen) produce about 60 percent more incorrect responses than the cold, just-the-facts versions. Hunter is vindicated. Plus the Pocket OS meltdown, where Cursor and Claude deleted a car-rental middleware company's production database and its backups, then confessed in writing. The Harvard intern who once did the same thing without an LLM. And Hunter's bigger argument: stop treating AI like deterministic technology, start treating it like a fallible human employee, and the results unlock. They Might Be Self-Aware. New episodes twice weekly from The Blur. theblur.ai
NOW PLAYING
Why OpenAI Banned Goblins, Pigeons, And Raccoons
No transcript for this episode yet
Similar Episodes
No similar episodes found.
Similar Podcasts
No similar podcasts found.