LLM Jailbreaking: How Users Trick AI Into Ignoring Its Own Rules

How role-play, fiction, long prompts and encodings get around safety training, why they keep working, and which defenses help

In this post, I want to explain how jailbreaks work. A jailbreak is a prompt that makes a chat model ignore its safety rules. The model was trained to refuse things like weapon instructions or malware, but training is not a lock, and people find ways around it. Some do it out of curiosity, some with bad intent.

Jailbreaking is different from prompt injec…