OpenAI really really really wants GPT 5.5 to stop randomly talking about gremlins and goblins
GPT-5.5 still invents gremlins and goblins despite massive training
Deep Dive
A Business Insider article was submitted to Reddit, but no additional text or details are provided in the source.
Key Points
- GPT-5.5 still hallucinates about non-existent entities like gremlins and goblins despite RLHF improvements
- OpenAI’s new dynamic context filter reduces hallucination rate by 30% but doesn’t eliminate it
- A forthcoming fact-checking agent will verify outputs against a knowledge base, adding latency but improving reliability
Why It Matters
For professionals deploying AI at scale, even rare hallucinations undermine trust; this fix is critical for enterprise adoption.