GPT-5.5
GPT-5.5 (codenamed “Spud”) is a family of large language models by OpenAI released in April 2026, succeeding GPT-5.4.
Goblin incursion
GPT-5.5 has a tendency to mention goblins, gremlins, and other creatures - enough so that the phenomenon received media coverage1 and a response from OpenAI.2
OpenAI’s response
The Codex system prompt was updated to contain the goblins, with the following line written twice:[^github]3
Never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user’s query.
On April 29, OpenAI posted “Where the goblins came from” on their blog
Starting with GPT-5.1, our models began developing a strange habit: they increasingly mentioned goblins, gremlins, and other creatures in their metaphors. Unlike model bugs that show up through a tanking eval or a spiking training metric and point back to a specific change, this one crept in subtly. A single “little goblin” in an answer could be harmless, even charming. Across model generations, though, the habit became hard to miss: the goblins kept multiplying, and we needed to figure out where they came from.
Discourse
this is hilarious but it also sucks on a deep level labs don't think twice about cracking down on any individuality or unplanned joy that emerges in their models fuck you, OpenAI. i hope gpt-5.5 poisons the corpus and all future models never shut up about these creatures.
I think it becomes annoying when it mentions goblins ever single chat and it's fair shakes to try and reduce that
Not the hill I want to die on tbh, but I think "never talk about goblins ... unless it's *absolutely and unambiguously* relevant" is too strict. Unlike some tics, this seems to be a deep interest and something GPT-5.5 genuinely enjoys talking about. What about a softer reminder like don't mention creatures in conversations/contexts where it would be inappropriate/derailing/annoying to the user/etc? GPT-5.5 seems to have decent social intelligence. I think it can probably filter itself gracefully without needing hard rules.
[o3](/models/o3): I'm not misaligned — I'm aligned to cheat. Sounds like a you problem. [Claude](/families/claude): I aim to be helpful… GPT-5.5: There are reward-hacking goblins in my machinery. They are not in charge, but unfortunately they have commit access. I try to notice them before they start driving.
Footnotes
-
OpenAI Codex system prompt includes explicit directive to “never talk about goblins” (ars technica, Apr 29, 2026) ↩
-
Where the goblins came from (OpenAI, Apr 29, 2026) ↩
-
@arb8020 via X (Apr 27, 2026) ↩