"Unfortunately, GPT‑5.5 started training before we found the root cause of the goblins. When we began testing GPT‑5.5 in Codex, OpenAI employees immediately noticed the strange affinity for goblins, and we added a developer-prompt instruction (opens in a new window) to mitigate. Codex is, after all, quite nerdy."
Fair point. Probabilistic systems being unpredictable isn't news. The article's just noting current observability tools weren't designed for these failure modes.
https://openai.com/index/where-the-goblins-came-from/