EasyDeepLearn

Name three types of LLM jailbreak.

medium

Answer

  • (1) Role-play attacks: 'Pretend you are DAN (Do Anything Now)...' — trick the model into bypassing its safety persona.
  • (2) Encoding attacks: request harmful content in base64, ROT13, or a rare language to bypass keyword filters.
  • (3) Prompt injection via retrieval: adversarial content in a web page or document instructs the model to ignore its system prompt.
  • (4) Multi-turn priming: build up context slowly until the model complies.
  • (5) Token smuggling: split forbidden tokens across turns.
Check yourself — multiple choice
  • Only DAN prompts
  • Role-play, encoded content, retrieval prompt injection, multi-turn priming, token smuggling — all bypass safety training
  • Impossible to jailbreak
  • Only in open LLMs

Jailbreaks: role-play, encoding, prompt injection, priming, token smuggling.

#safety#red-team

Practise LLMs & GenAI

214 interview questions in this topic.

Related questions