Why OpenAI's 'goblin' problem matters — and how you can release the goblins on your own

OpenAI said GPT-5.5’s odd “goblin” behavior came from its built-in personality customization system, not a bug.

A viral find in the Codex GitHub repository showed instructions telling the model not to mention goblins and similar creatures.

The discovery sparked speculation on X, Reddit, and Hacker News about how ChatGPT and GPT-5.5 are shaped by prompts and RLHF.

The case underscores how model behavior can be influenced by training and product design, with implications for trust and developer control.