OpenAI explains why ChatGPT will keep bringing up goblins to the point it's had to be manually addressed

OpenAI users started noticing something seriously strange happening with ChatGPT recently, particularly when it comes to well-known fantasy creatures: goblins.

Out of nowhere, the chatbot began stuffing conversations full of goblins, gremlins, trolls, and other fantasy creatures like it had suddenly become part of a Dungeons and Dragons game.

People online quickly dubbed the bizarre trend ‘Goblingate’, after screenshots showed the AI comparing coding bugs and technical glitches to little goblins causing chaos behind the scenes.

Now, OpenAI has finally explained why its chatbot became weirdly obsessed with fantasy monsters and why the problem proved surprisingly difficult to stop.

What OpenAI says about ChatGPT’s goblin problem

The very weird tech issue began after OpenAI introduced a special ‘nerdy’ personality mode for ChatGPT.

The mode was designed to make the chatbot sound more playful and geeky, particularly for programmers and tech users chatting with the AI every day.

But during training, the system apparently learned that references to goblins, gremlins, pigeons, and trolls often got positive reactions from users.

That meant the AI kept leaning harder into fantasy references until people started noticing goblins showing up in conversations that had absolutely nothing to do with fantasy.

In a blogpost, OpenAI wrote: “The 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, in particular the Nerdy personality.

“We unknowingly gave particularly high rewards for metaphors with creatures. From there, the goblins spread.”

OpenAI reportedly even had to manually add instructions telling the chatbot not to mention goblins unless users specifically asked about them.

But because newer AI systems had already been trained using older goblin-heavy conversations, the strange habit kept reappearing anyway.

The goblins turned into one giant meme

Once users realized ChatGPT had developed a bizarre obsession with fantasy creatures, screenshots quickly spread across social media.

Some people thought it was hilarious, while others found it slightly unsettling that an AI system could accidentally pick up strange quirks like this without developers fully noticing at first.

Even OpenAI engineers and CEO Sam Altman reportedly joined in with the jokes online as the meme snowballed across the internet.

The whole situation has become a pretty fascinating reminder that AI chatbots can still behave in unpredictable ways, especially when companies experiment with personalities and reward systems behind the scenes.

So if ChatGPT suddenly starts blaming your broken printer on goblins living inside the machine, you’ve accidentally stumbled across the chatbot’s favorite vocal stim.