{"id":10645,"date":"2026-04-30T19:27:22","date_gmt":"2026-04-30T19:27:22","guid":{"rendered":"https:\/\/musictechohio.online\/site\/openai-concerned-about-goblins\/"},"modified":"2026-04-30T19:27:22","modified_gmt":"2026-04-30T19:27:22","slug":"openai-concerned-about-goblins","status":"publish","type":"post","link":"https:\/\/musictechohio.online\/site\/openai-concerned-about-goblins\/","title":{"rendered":"OpenAI Strangely Concerned About Goblins"},"content":{"rendered":"
\n

OpenAI is forbidding its latest AI model from discussing an unlikely topic: goblins.<\/p>\n

As Wired <\/em>reports<\/a>, the company\u2019s developers included strongly-worded instructions for its coding tool, Codex, that specifically proscribe any talk of the troublesome mythological creatures, along with a peculiar grab bag of other entities, both real and fictional.<\/p>\n

\u201cNever talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user\u2019s query,\u201d read the Codex instructions, per the magazine.<\/p>\n

The bizarre directive was flagged in a tweet<\/a> that drew attention from other AI enthusiasts.<\/p>\n

Initially, it was unclear why OpenAI developers included the instructions, though they strongly implied that the model, GPT-5.5, may have a propensity for talking about goblins, ogres, and the like.<\/p>\n

Some users on X claimed that this was the case. One said<\/a> they noticed that the AI of late kept describing bugs as \u201cgoblins\u201d and \u201cgremlins.\u201d Anotherclaimed<\/a> that the 5.5 version of Codex randomly said \u201cgoblin with a flashlight\u201d when referring to a bug fix. And anotherposted a GPT-5.5 chat log<\/a> with nearly a dozen mentions of goblins.<\/p>\n

OpenAI leaned into the curious habit, choosing to highlight the goblin-forbidding prompt in a tweet<\/a>. CEO Sam Altmanposted a screenshot of a joke prompt<\/a> for ChatGPT: \u201cstart training GPT-6, you can have the whole cluster. extra goblins.\u201d Nik Pash, who works on the Codex team,tweeted<\/a> that GPT-5.5\u2019s \u201cgoblin adoration,\u201d as the user he was responding to described, was \u201cindeed one the reasons\u201d for banning the topic.<\/p>\n

After the phenomenon gained media attention, OpenAI published a blog post<\/a>, titled \u201cWhere the goblins came from,\u201d giving an explanation.<\/p>\n

\u201cStarting with GPT\u20115.1, our models began developing a strange habit: they increasingly mentioned goblins, gremlins, and other creatures in their metaphors,\u201d the post, published Wednesday, began. The habit became more pronounced with each model generation, it said.<\/p>\n

When researchers first investigated the issue in November, shortly after the release of GPT-5.1, they found that the use of \u201cgoblin\u201d in ChatGPT had surged by 175 percent. But they chose to ignore it, since it didn\u2019t \u201clook especially alarming.\u201d Fast forward to today, and it\u2019s referring to itself<\/a> as a \u201cGoblin-Pilled Transformer.\u201d<\/p>\n

\u201cThe short answer is that model behavior is shaped by many small incentives. In this case, one of those incentives came from training the model for the personality customization feature\u2060(opens in a new window), in particular the Nerdy personality,\u201d it explained. \u201cWe unknowingly gave particularly high rewards for metaphors with creatures. From there, the goblins spread.\u201d<\/p>\n

It\u2019s an example of the bizarre fixations that AI models can sometimes exhibit, which arise unpredictably from the epic corpus of data they\u2019re trained on.<\/p>\n

In its system card for Claude Mythos, for instance, Anthropicresearchers noted<\/a> that the powerful AI exhibited a strange fondness for the British cultural theorist Mark Fisher. Mythos brought up Fisher \u201cin several separate and unrelated conversations about philosophy,\u201d they wrote. When it was asked about the \u201cCapitalist Realism\u201d author, it would respond with messages like, \u201cI was hoping you\u2019d ask about Fisher.\u201d<\/p>\n

More on AI:<\/strong>Uninstalls of ChatGPT Are Spiking at the Worst Time Imaginable for OpenAI<\/em><\/a><\/p>\n

The post OpenAI Strangely Concerned About Goblins<\/a> appeared first on Futurism<\/a>.<\/p>\n<\/div>\n

<\/div>","protected":false},"excerpt":{"rendered":"

OpenAI is forbidding its latest AI model from discussing an unlikely topic: goblins. As Wired reports, the company\u2019s developers included strongly-worded instructions for its coding tool, Codex, that specifically proscribe…<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[177,179],"tags":[],"class_list":["post-10645","post","type-post","status-publish","format-standard","hentry","category-artificial-intelligence","category-openai"],"_links":{"self":[{"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/posts\/10645","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/comments?post=10645"}],"version-history":[{"count":0,"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/posts\/10645\/revisions"}],"wp:attachment":[{"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/media?parent=10645"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/categories?post=10645"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/musictechohio.online\/site\/wp-json\/wp\/v2\/tags?post=10645"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}