OpenAI reveals training bug in Nerdy personality that spreads creature metaphors across GPT models. Read all about the ...
ChatGPT’s unusual fixation on goblins highlighted how subtle AI design choices can create unintended behaviours, prompting ...
OpenAI admits a personality training flaw caused ChatGPT to repeatedly use “goblin” references across GPT models and Codex.
OpenAI has solved a “goblin mystery” impacting ChatGPT that caused the AI chatbot to become obsessed with the mythical ...
In a somewhat unusual post earlier this week, ChatGPT makers OpenAI said that starting with GPT-5.1, their models had ...
OpenAI has told its ChatGPT and Codex models to avoid mentioning goblins, gremlins, and similar creatures unless directly relevant, after a surge in unrelated references appeared in outputs.
OpenAI did something pretty unusual. Engineers went into ChatGPT's production code and added a rule: never mention goblins.
Hosted on MSN
Sam Altman teases GPT-6 amid quirky 'goblin' glitch
OpenAI CEO Sam Altman has hinted that GPT-6 is in development, making the remark in the wake of a humorous 'goblin' glitch in its Codex tool. While GPT-5.5 was released only recently, Altman’s comment ...
A surprising change in OpenAI's tools has caught the attention of developers and researchers. The company instructed its ...
OpenAI traced ChatGPT’s excessive goblin and gremlin references to reinforcement learning that over-rewarded creative ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results